Authors: Keivan Alizadeh, Iman Mirzadeh, Hooman Shahrokhi, Dmitry Belenko, Frank Sun, Minsik Cho, Mohammad Hossein Sekhavat, Moin Nabi, Mehrdad Farajtabar
Abstract: Large Language Models (LLMs) typically generate outputs token by token using a fixed compute budget, leading to inefficient resource utilization. To address this shortcoming, recent advancements in mixture of expert (MoE) models, speculative decoding, and early exit strategies leverage the insight that computational demands can vary significantly based on the complexity and nature of the input. However, identifying optimal routing patterns for dynamic execution remains an open challenge, limiting the full potential of these adaptive methods. To address this need, we study adaptive computation in LLMs more systematically. We propose a novel framework that integrates smaller auxiliary modules within each Feed-Forward Network layer of the LLM. This design enables dynamic routing of tokens based on task complexity: tokens can be processed by either the small or big modules at each layer, or even bypass certain layers entirely. This allows us to introduce a novel notion of a token's difficulty, defined by its potential to benefit from additional computational resources. Importantly, by employing oracles to identify optimal patterns of adaptive computations, we gain valuable insights into the internal workings of LLMs and the routing processes in a simplified heterogeneous MoE setup. We show that trained routers operate differently from oracles and often yield suboptimal solutions. Notably, activating a large module in just one layer outperforms models that use large modules across all layers, underscoring the gap between practical implementations of routing in MoE models and theoretical optima for adaptive computation.
Authors: He Li, Jianhang Hong, Yuanzhuo Wu, Snehal Adbol, Zonglin Li
Abstract: Model compression methods are used to reduce the computation and energy requirements for Large Language Models (LLMs). Quantization Aware Training (QAT), an effective model compression method, is proposed to reduce performance degradation after quantization. To further minimize this degradation, we introduce two continuous approximations to the QAT process on the rounding function, traditionally approximated by the Straight-Through Estimator (STE), and the clamping function. By applying both methods, the perplexity (PPL) on the WikiText-v2 dataset of the quantized model reaches 9.0815, outperforming 9.9621 by the baseline. Also, we achieve a 2.76% improvement on BoolQ, and a 5.47% improvement on MMLU, proving that the step sizes and weights can be learned more accurately with our approach. Our method achieves better performance with the same precision, model size, and training setup, contributing to the development of more energy-efficient LLMs technology that aligns with global sustainability goals.
Authors: Jingyang Qiao, Zhizhong Zhang, Xin Tan, Yanyun Qu, Shouhong Ding, Yuan Xie
Abstract: Instruction tuning guides the Multimodal Large Language Models (MLLMs) in aligning different modalities by designing text instructions, which seems to be an essential technique to enhance the capabilities and controllability of foundation models. In this framework, Multimodal Continual Instruction Tuning (MCIT) is adopted to continually instruct MLLMs to follow human intent in sequential datasets. We observe existing gradient update would heavily destroy the tuning performance on previous datasets and the zero-shot ability during continual instruction tuning. Exponential Moving Average (EMA) update policy owns the ability to trace previous parameters, which can aid in decreasing forgetting. However, its stable balance weight cannot deal with the ever-changing datasets, leading to the out-of-balance between plasticity and stability of MLLMs. In this paper, we propose a method called Multimodal Large Language Continual Assistant (LLaCA) to address the challenge. Starting from the trade-off prerequisite and EMA update, we propose the plasticity and stability ideal condition. Based on Taylor expansion in the loss function, we find the optimal balance weight is basically according to the gradient information and previous parameters. We automatically determine the balance weight and significantly improve the performance. Through comprehensive experiments on LLaVA-1.5 in a continual visual-question-answering benchmark, compared with baseline, our approach not only highly improves anti-forgetting ability (with reducing forgetting from 22.67 to 2.68), but also significantly promotes continual tuning performance (with increasing average accuracy from 41.31 to 61.89). Our code will be published soon.
Authors: Mingliang Liang, Martha Larson
Abstract: We propose Word-Frequency-based Image-Text Pair Pruning (WFPP), a novel data pruning method that improves the efficiency of VLMs. Unlike MetaCLIP, our method does not need metadata for pruning, but selects text-image pairs to prune based on the content of the text. Specifically, WFPP prunes text-image pairs containing high-frequency words across the entire training dataset. The effect of WFPP is to reduce the dominance of frequent words. The result a better balanced word-frequency distribution in the dataset, which is known to improve the training of word embedding models. After pre-training on the pruned subset, we fine-tuned the model on the entire dataset for one additional epoch to achieve better performance. Our experiments demonstrate that applying WFPP when training a CLIP model improves performance on a wide range of downstream tasks. WFPP also provides the advantage of speeding up pre-training by using fewer samples. Additionally, we analyze the training data before and after pruning to visualize how WFPP changes the balance of word frequencies. We hope our work encourages researchers to consider the distribution of words in the training data when pre-training VLMs, not limited to CLIP.
Authors: Sudhakar Sah, Ravish Kumar, Darshan C. Ganji, Ehsan Saboori
Abstract: Activation functions introduce non-linearity into Neural Networks, enabling them to learn complex patterns. Different activation functions vary in speed and accuracy, ranging from faster but less accurate options like ReLU to slower but more accurate functions like SiLU or SELU. Typically, same activation function is used throughout an entire model architecture. In this paper, we conduct a comprehensive study on the effects of using mixed activation functions in YOLO-based models, evaluating their impact on latency, memory usage, and accuracy across CPU, NPU, and GPU edge devices. We also propose a novel approach that leverages Neural Architecture Search (NAS) to design YOLO models with optimized mixed activation functions.The best model generated through this method demonstrates a slight improvement in mean Average Precision (mAP) compared to baseline model (SiLU), while it is 22.28% faster and consumes 64.15% less memory on the reference NPU device.
Authors: Xurui Li, Juanjuan Yao
Abstract: The advent of Large Language Models (LLMs) has ushered in a new era of artificial intelligence, with the potential to transform various sectors through automation and insightful analysis. The Mixture of Experts (MoE) architecture has been proposed as a solution to enhance model performance in complex tasks. Yet, existing MoE models struggle with task-specific learning and interpretability, especially in fields like medicine where precision is critical. This paper introduces the Adaptive Task-planing Mixture of Experts(AT-MoE), an innovative architecture designed to address these limitations. We first train task-specific experts via LoRA approach to enhance problem-solving capabilities and interpretability in specialized areas. Subsequently, we introduce a layer-wise adaptive grouped routing module that optimizes module fusion based on complex task instructions, ensuring optimal task resolution. The grouped routing module first perform overall weight allocation from the dimension of the expert group, and then conduct local weight normalization adjustments within the group. This design maintains multi-dimensional balance, controllability, and interpretability, while facilitating task-specific fusion in response to complex instructions.
Authors: Hongxin Ding, Yue Fang, Runchuan Zhu, Xinke Jiang, Jinyang Zhang, Yongxin Xu, Xu Chu, Junfeng Zhao, Yasha Wang
Abstract: Large Language Models(LLMs) excel in general tasks but struggle in specialized domains like healthcare due to limited domain-specific knowledge.Supervised Fine-Tuning(SFT) data construction for domain adaptation often relies on heuristic methods, such as GPT-4 annotation or manual data selection, with a data-centric focus on presumed diverse, high-quality datasets. However, these methods overlook the model's inherent knowledge distribution, introducing noise, redundancy, and irrelevant data, leading to a mismatch between the selected data and the model's learning task, resulting in suboptimal performance. To address this, we propose a two-stage model-centric data selection framework, Decomposed Difficulty Data Selection (3DS), which aligns data with the model's knowledge distribution for optimized adaptation. In Stage1, we apply Prompt-Driven Data Selection via Explicit Alignment, where the the model filters irrelevant or redundant data based on its internal knowledge. In Stage2, we perform Decomposed Difficulty Data Selection, where data selection is guided by our defined difficulty decomposition, using three metrics: Instruction Understanding, Response Confidence, and Response Correctness. Additionally, an attention-based importance weighting mechanism captures token importance for more accurate difficulty calibration. This two-stage approach ensures the selected data is not only aligned with the model's knowledge and preferences but also appropriately challenging for the model to learn, leading to more effective and targeted domain adaptation. In the case study of the medical domain, our extensive experiments on real-world healthcare datasets demonstrate the superiority of 3DS over exisiting methods in accuracy by over 5.29%. Our dataset and code will be open-sourced at https://anonymous.4open.science/r/3DS-E67F.
Authors: Andrew Jesson, Yiding Jiang
Abstract: We demonstrate that recent advances in reinforcement learning (RL) combined with simple architectural changes significantly improves generalization on the ProcGen benchmark. These changes are frame stacking, replacing 2D convolutional layers with 3D convolutional layers, and scaling up the number of convolutional kernels per layer. Experimental results using a single set of hyperparameters across all environments show a 37.9\% reduction in the optimality gap compared to the baseline (from 0.58 to 0.36). This performance matches or exceeds current state-of-the-art methods. The proposed changes are largely orthogonal and therefore complementary to the existing approaches for improving generalization in RL, and our results suggest that further exploration in this direction could yield substantial improvements in addressing generalization challenges in deep reinforcement learning.
Authors: Mohammad Al-Sayed Ahmad, Jude Haddad
Abstract: Thyroid carcinoma, a significant yet often controllable cancer, has seen a rise in cases, largely due to advancements in diagnostic methods. Differentiated thyroid cancer (DTC), which includes papillary and follicular varieties, is typically associated with a positive prognosis in academic circles. Nevertheless, there are still some individuals who may experience a recurrence. This study employs machine learning, particularly deep learning models, to predict the recurrence of DTC, with the goal of improving patient care through personalized treatment approaches. By analysing a dataset containing clinicopathological features of patients, the model achieved remarkable accuracy rates of 98% during training and 96% during testing. To improve the model's interpretability, we used techniques like LIME and Morris Sensitivity Analysis. These methods gave us valuable insights into how the model makes decisions. The results suggest that combining deep learning models with interpretability techniques can be extremely useful in quickly identifying the recurrence of thyroid cancer in patients. This can help in making informed therapeutic choices and customizing treatment approaches for individual patients.
Authors: Edward Berman, Jacob Ginesin
Abstract: Julia has been heralded as a potential successor to Python for scientific machine learning and numerical computing, boasting ergonomic and performance improvements. Since Julia's inception in 2012 and declaration of language goals in 2017, its ecosystem and language-level features have grown tremendously. In this paper, we take a modern look at Julia's features and ecosystem, assess the current state of the language, and discuss its viability and pitfalls as a replacement for Python as the de-facto scientific machine learning language. We call for the community to address Julia's language-level issues that are preventing further adoption.
Authors: Haiquan Lu, Yefan Zhou, Shiwei Liu, Zhangyang Wang, Michael W. Mahoney, Yaoqing Yang
Abstract: Recent work on pruning large language models (LLMs) has shown that one can eliminate a large number of parameters without compromising performance, making pruning a promising strategy to reduce LLM model size. Existing LLM pruning strategies typically assign uniform pruning ratios across layers, limiting overall pruning ability; and recent work on layerwise pruning of LLMs is often based on heuristics that can easily lead to suboptimal performance. In this paper, we leverage Heavy-Tailed Self-Regularization (HT-SR) Theory, in particular the shape of empirical spectral densities (ESDs) of weight matrices, to design improved layerwise pruning ratios for LLMs. Our analysis reveals a wide variability in how well-trained, and thus relatedly how prunable, different layers of an LLM are. Based on this, we propose AlphaPruning, which uses shape metrics to allocate layerwise sparsity ratios in a more theoretically principled manner. AlphaPruning can be used in conjunction with multiple existing LLM pruning methods. Our empirical results show that AlphaPruning prunes LLaMA-7B to 80% sparsity while maintaining reasonable perplexity, marking a first in the literature on LLMs. We have open-sourced our code at https://github.com/haiquanlu/AlphaPruning.
Authors: Shen Yuan, Hongteng Xu
Abstract: Transformer plays a central role in many fundamental deep learning models, e.g., the ViT in computer vision and the BERT and GPT in natural language processing, whose effectiveness is mainly attributed to its multi-head attention (MHA) mechanism. In this study, we propose a simple and novel channel-wise sample permutation (CSP) operator, achieving a new structured MHA with fewer parameters and lower complexity. Given an input matrix, CSP circularly shifts the samples of different channels with various steps and then sorts grouped samples of each channel. This operator is equivalent to implicitly implementing cross-channel attention maps as permutation matrices, which achieves linear complexity and suppresses the risk of rank collapse when representing data. We replace the MHA of some representative models with CSP and test the CSP-based models in several discriminative tasks, including image classification and long sequence analysis. Experiments show that the CSP-based models achieve comparable or better performance with fewer parameters and lower computational costs than the classic Transformer and its state-of-the-art variants. The code is available at https://github.com/DaShenZi721/CSP.
Authors: Qianru Zhang, Haixin Wang, Siu-Ming Yiu, Hongzhi Yin
Abstract: Effective spatio-temporal prediction frameworks play a crucial role in urban sensing applications, including traffic analysis, human mobility behavior modeling, and citywide crime prediction. However, the presence of data noise and label sparsity in spatio-temporal data presents significant challenges for existing neural network models in learning effective and robust region representations. To address these challenges, we propose a novel spatio-temporal graph masked autoencoder paradigm that explores generative self-supervised learning for effective spatio-temporal data augmentation. Our proposed framework introduces a spatial-temporal heterogeneous graph neural encoder that captures region-wise dependencies from heterogeneous data sources, enabling the modeling of diverse spatial dependencies. In our spatio-temporal self-supervised learning paradigm, we incorporate a masked autoencoding mechanism on node representations and structures. This mechanism automatically distills heterogeneous spatio-temporal dependencies across regions over time, enhancing the learning process of dynamic region-wise spatial correlations. To validate the effectiveness of our STGMAE framework, we conduct extensive experiments on various spatio-temporal mining tasks. We compare our approach against state-of-the-art baselines. The results of these evaluations demonstrate the superiority of our proposed framework in terms of performance and its ability to address the challenges of spatial and temporal data noise and sparsity in practical urban sensing scenarios.
Authors: Liubov Tupikina (UPD5, LPI), Kathuria Hritika (LPI)
Abstract: Active area of research in AI is the theory of manifold learning and finding lower-dimensional manifold representation on how we can learn geometry from data for providing better quality curated datasets. There are however various issues with these methods related to finding low-dimensional representation of the data, the so-called curse of dimensionality. Geometric deep learning methods for data learning often include set of assumptions on the geometry of the feature space. Some of these assumptions include pre-selected metrics on the feature space, usage of the underlying graph structure, which encodes the data points proximity. However, the later assumption of using a graph as the underlying discrete structure, encodes only the binary pairwise relations between data points, restricting ourselves from capturing more complex higher-order relationships, which are often often present in various systems. These assumptions together with data being discrete and finite can cause some generalisations, which are likely to create wrong interpretations of the data and models outputs. Hence overall this can cause wrong outputs of the embedding models themselves, while these models being quite and trained on large corpora of data, such as BERT, Yi and other similar models.The objective of our research is twofold, first, it is to develop the alternative framework to characterize the embedding methods dissecting their possible inconsistencies using combinatorial approach of higher-order structures which encode the embedded data. Second objective is to explore the assumption of the underlying structure of embeddings to be graphs, substituting it with the hypergraph and using the hypergraph theory to analyze this structure. We also demonstrate the embedding characterization on the usecase of the arXiv data.
Authors: Hanlin Gu, Hong Xi Tae, Chee Seng Chan, Lixin Fan
Abstract: This paper addresses the critical challenge of unlearning in Vertical Federated Learning (VFL), an area that has received limited attention compared to horizontal federated learning. We introduce the first approach specifically designed to tackle label unlearning in VFL, focusing on scenarios where the active party aims to mitigate the risk of label leakage. Our method leverages a limited amount of labeled data, utilizing manifold mixup to augment the forward embedding of insufficient data, followed by gradient ascent on the augmented embeddings to erase label information from the models. This combination of augmentation and gradient ascent enables high unlearning effectiveness while maintaining efficiency, completing the unlearning procedure within seconds. Extensive experiments conducted on diverse datasets, including MNIST, CIFAR10, CIFAR100, and ModelNet, validate the efficacy and scalability of our approach. This work represents a significant advancement in federated learning, addressing the unique challenges of unlearning in VFL while preserving both privacy and computational efficiency.
Authors: Hong Li, Zhiquan Tan, Xingyu Li, Weiran Huang
Abstract: While vision-and-language models significantly advance in many fields, the challenge of continual learning is unsolved. Parameter-efficient modules like adapters and prompts present a promising way to alleviate catastrophic forgetting. However, existing works usually learn individual adapters for each task, which may result in redundant knowledge among adapters. Moreover, they continue to use the original pre-trained model to initialize the downstream model, leading to negligible changes in the model's generalization compared to the original model. In addition, there is still a lack of research investigating the consequences of integrating a multi-modal model into the updating procedure for both uni-modal and multi-modal tasks and the subsequent impacts it has on downstream tasks. In this paper, we propose an adapter-based two-stage learning paradigm, a multi-modal continual learning scheme that consists of experience-based learning and novel knowledge expansion, which helps the model fully use experience knowledge and compensate for novel knowledge. Extensive experiments demonstrate that our method is proficient for continual learning. It expands the distribution of representation upstream while also minimizing the negative impact of forgetting previous tasks. Additionally, it enhances the generalization capability for downstream tasks. Furthermore, we incorporate both multi-modal and uni-modal tasks into upstream continual learning. We observe that learning from upstream tasks can help with downstream tasks. Our code will be available at: https://github.com/lihong2303/ATLAS.
Authors: Zhen Qin, Zhaomin Wu, Bingsheng He, Shuiguang Deng
Abstract: Instruction tuning helps improve pretrained large language models (LLMs) in terms of the responsiveness to human instructions, which is benefited from diversified instruction data. Federated learning extends the sources of instruction data by exploiting the diversified client-side data, making it increasingly popular for tuning LLMs. Existing approaches of federated LLM tuning typically traverse all local data during local training, bringing excessive computation overhead and posing a risk of overfitting local data. Thus, a federated data-efficient instruction tuning approach, which consumes relatively little data from the entire dataset, is needed. In response, this work introduces an approach of federated data-efficient instruction tuning for LLMs, FedHDS, which utilizes a representative subset of edge-side data, coreset, to tune the LLM. It reduces the redundancy of data samples at both intra-client and inter-client levels through a hierarchical data selection framework performed by jointly selecting a small number of representative data samples for local training without sharing the raw data. Extensive experiments conducted across six scenarios with various LLMs, datasets and data partitions demonstrate that FedHDS significantly reduces the amount of data required for fine-tuning while improving the responsiveness of the instruction-tuned LLMs to unseen tasks.
Authors: Christofel Rio Goenawan, Har Dong-Soo
Abstract: In the modern world, the development of Artificial Intelligence (AI) has contributed to improvements in various areas, including automation, computer vision, fraud detection, and more. AI can be leveraged to enhance the efficiency of Autonomous Smart Traffic Management (ASTM) systems and reduce traffic congestion rates. This paper presents an Autonomous Smart Traffic Management (STM) system that uses AI to improve traffic flow rates. The system employs the YOLO V5 Convolutional Neural Network to detect vehicles in traffic management images. Additionally, it predicts the number of vehicles for the next 12 hours using a Recurrent Neural Network with Long Short-Term Memory (RNN-LSTM). The Smart Traffic Management Cycle Length Analysis manages the traffic cycle length based on these vehicle predictions, aided by AI. From the results of the RNN-LSTM model for predicting vehicle numbers over the next 12 hours, we observe that the model predicts traffic with a Mean Squared Error (MSE) of 4.521 vehicles and a Root Mean Squared Error (RMSE) of 2.232 vehicles. After simulating the STM system in the CARLA simulation environment, we found that the Traffic Management Congestion Flow Rate with ASTM (21 vehicles per minute) is 50\% higher than the rate without STM (around 15 vehicles per minute). Additionally, the Traffic Management Vehicle Pass Delay with STM (5 seconds per vehicle) is 70\% lower than without STM (around 12 seconds per vehicle). These results demonstrate that the STM system using AI can increase traffic flow by 50\% and reduce vehicle pass delays by 70\%.
Authors: Shiran Yuan, Hao Zhao
Abstract: We address an important problem in ecology called Species Distribution Modeling (SDM), whose goal is to predict whether a species exists at a certain position on Earth. In particular, we tackle a challenging version of this task, where we learn from presence-only data in a community-sourced dataset, model a large number of species simultaneously, and do not use any additional environmental information. Previous work has used neural implicit representations to construct models that achieve promising results. However, implicit representations often generate predictions of limited spatial precision. We attribute this limitation to their inherently global formulation and inability to effectively capture local feature variations. This issue is especially pronounced with presence-only data and a large number of species. To address this, we propose a hybrid embedding scheme that combines both implicit and explicit embeddings. Specifically, the explicit embedding is implemented with a multiresolution hashgrid, enabling our models to better capture local information. Experiments demonstrate that our results exceed other works by a large margin on various standard benchmarks, and that the hybrid representation is better than both purely implicit and explicit ones. Qualitative visualizations and comprehensive ablation studies reveal that our hybrid representation successfully addresses the two main challenges. Our code is open-sourced at https://github.com/Shiran-Yuan/HSR-SDM.
Authors: Farhang Yeganegi, Arian Eamaz, Mojtaba Soltanalian
Abstract: Deep learning models excel at capturing complex representations through sequential layers of linear and non-linear transformations, yet their inherent black-box nature and multi-modal training landscape raise critical concerns about reliability, robustness, and safety, particularly in high-stakes applications. To address these challenges, we introduce YES training bounds, a novel framework for real-time, data-aware certification and monitoring of neural network training. The YES bounds evaluate the efficiency of data utilization and optimization dynamics, providing an effective tool for assessing progress and detecting suboptimal behavior during training. Our experiments show that the YES bounds offer insights beyond conventional local optimization perspectives, such as identifying when training losses plateau in suboptimal regions. Validated on both synthetic and real data, including image denoising tasks, the bounds prove effective in certifying training quality and guiding adjustments to enhance model performance. By integrating these bounds into a color-coded cloud-based monitoring system, we offer a powerful tool for real-time evaluation, setting a new standard for training quality assurance in deep learning.
Authors: Weronika Ormaniec, Felix Dangel, Sidak Pal Singh
Abstract: The Transformer architecture has inarguably revolutionized deep learning, overtaking classical architectures like multi-layer perceptrons (MLPs) and convolutional neural networks (CNNs). At its core, the attention block differs in form and functionality from most other architectural components in deep learning -- to the extent that Transformers are often accompanied by adaptive optimizers, layer normalization, learning rate warmup, and more, in comparison to MLPs/CNNs. The root causes behind these outward manifestations, and the precise mechanisms that govern them, remain poorly understood. In this work, we bridge this gap by providing a fundamental understanding of what distinguishes the Transformer from the other architectures -- grounded in a theoretical comparison of the (loss) Hessian. Concretely, for a single self-attention layer, (a) we first entirely derive the Transformer's Hessian and express it in matrix derivatives; (b) we then characterize it in terms of data, weight, and attention moment dependencies; and (c) while doing so further highlight the important structural differences to the Hessian of classical networks. Our results suggest that various common architectural and optimization choices in Transformers can be traced back to their highly non-linear dependencies on the data and weight matrices, which vary heterogeneously across parameters. Ultimately, our findings provide a deeper understanding of the Transformer's unique optimization landscape and the challenges it poses.
Authors: Byron (Pin-Lun), Hsu, Yun Dai, Vignesh Kothapalli, Qingquan Song, Shao Tang, Siyu Zhu, Steven Shimizu, Shivam Sahni, Haowen Ning, Yanning Chen
Abstract: Training Large Language Models (LLMs) efficiently at scale presents a formidable challenge, driven by their ever-increasing computational demands and the need for enhanced performance. In this work, we introduce Liger-Kernel, an open-sourced set of Triton kernels developed specifically for LLM training. With kernel optimization techniques like kernel operation fusing and input chunking, our kernels achieve on average a 20% increase in training throughput and a 60% reduction in GPU memory usage for popular LLMs compared to HuggingFace implementations. In addition, Liger-Kernel is designed with modularity, accessibility, and adaptability in mind, catering to both casual and expert users. Comprehensive benchmarks and integration tests are built in to ensure compatibility, performance, correctness, and convergence across diverse computing environments and model architectures. The source code is available under a permissive license at: github.com/linkedin/Liger-Kernel.
Authors: Harley Wiltzer, Marc G. Bellemare, David Meger, Patrick Shafto, Yash Jhaveri
Abstract: When decisions are made at high frequency, traditional reinforcement learning (RL) methods struggle to accurately estimate action values. In turn, their performance is inconsistent and often poor. Whether the performance of distributional RL (DRL) agents suffers similarly, however, is unknown. In this work, we establish that DRL agents are sensitive to the decision frequency. We prove that action-conditioned return distributions collapse to their underlying policy's return distribution as the decision frequency increases. We quantify the rate of collapse of these return distributions and exhibit that their statistics collapse at different rates. Moreover, we define distributional perspectives on action gaps and advantages. In particular, we introduce the superiority as a probabilistic generalization of the advantage -- the core object of approaches to mitigating performance issues in high-frequency value-based RL. In addition, we build a superiority-based DRL algorithm. Through simulations in an option-trading domain, we validate that proper modeling of the superiority distribution produces improved controllers at high decision frequencies.
Authors: Michael Painter
Abstract: In this paper, we develop novel techniques that can be used to alter the architecture of a neural network, while maintaining the function it represents. Such operations are known as function preserving transforms and have proven useful in transferring knowledge between networks to evaluate architectures quickly, thus having applications in efficient architectures searches. Our methods allow the integration of residual connections into function preserving transforms, so we call them R2R. We provide a derivation for R2R and show that it yields competitive performance with other function preserving transforms, thereby decreasing the restrictions on deep learning architectures that can be extended through function preserving transforms. We perform a comparative analysis with other function preserving transforms such as Net2Net and Network Morphisms, where we shed light on their differences and individual use cases. Finally, we show the effectiveness of R2R to train models quickly, as well as its ability to learn a more diverse set of filters on image classification tasks compared to Net2Net and Network Morphisms.
Authors: Bo Tang, Elias B. Khalil, J\'an Drgo\v{n}a
Abstract: Mixed-integer non-linear programs (MINLPs) arise in various domains, such as energy systems and transportation, but are notoriously difficult to solve. Recent advances in machine learning have led to remarkable successes in optimization tasks, an area broadly known as learning to optimize. This approach includes using predictive models to generate solutions for optimization problems with continuous decision variables, thereby avoiding the need for computationally expensive optimization algorithms. However, applying learning to MINLPs remains challenging primarily due to the presence of integer decision variables, which complicate gradient-based learning. To address this limitation, we propose two differentiable correction layers that generate integer outputs while preserving gradient information. Combined with a soft penalty for constraint violation, our framework can tackle both the integrality and non-linear constraints in a MINLP. Experiments on three problem classes with convex/non-convex objective/constraints and integer/mixed-integer variables show that the proposed learning-based approach consistently produces high-quality solutions for parametric MINLPs extremely quickly. As problem size increases, traditional exact solvers and heuristic methods struggle to find feasible solutions, whereas our approach continues to deliver reliable results. Our work extends the scope of learning-to-optimize to MINLP, paving the way for integrating integer constraints into deep learning models. Our code is available at https://github.com/pnnl/L2O-pMINLP.
Authors: Dhruva Chayapathy, Tavis Siebert, Lucas Spangher, Akshata Kishore Moharir, Om Manoj Patil, Cristina Rea
Abstract: Machine Learning guided data augmentation may support the development of technologies in the physical sciences, such as nuclear fusion tokamaks. Here we endeavor to study the problem of detecting disruptions i.e. plasma instabilities that can cause significant damages, impairing the reliability and efficiency required for their real world viability. Machine learning (ML) prediction models have shown promise in detecting disruptions for specific tokamaks, but they often struggle in generalizing to the diverse characteristics and dynamics of different machines. This limits the effectiveness of ML models across different tokamak designs and operating conditions, which is a critical barrier to scaling fusion technology. Given the success of data augmentation in improving model robustness and generalizability in other fields, this study explores the use of a novel time series viewmaker network to generate diverse augmentations or "views" of training data. Our results show that incorporating views during training improves AUC and F2 scores on DisruptionBench tasks compared to standard or no augmentations. This approach represents a promising step towards developing more broadly applicable ML models for disruption avoidance, which is essential for advancing fusion technology and, ultimately, addressing climate change through reliable and sustainable energy production.
Authors: Szymon Mi{\l}osz, Pawe{\l} Kapusta
Abstract: This study addresses the challenge of quantifying chess puzzle difficulty - a complex task that combines elements of game theory and human cognition and underscores its critical role in effective chess training. We present GlickFormer, a novel transformer-based architecture that predicts chess puzzle difficulty by approximating the Glicko-2 rating system. Unlike conventional chess engines that optimize for game outcomes, GlickFormer models human perception of tactical patterns and problem-solving complexity. The proposed model utilizes a modified ChessFormer backbone for spatial feature extraction and incorporates temporal information via factorized transformer techniques. This approach enables the capture of both spatial chess piece arrangements and move sequences, effectively modeling spatio-temporal relationships relevant to difficulty assessment. Experimental evaluation was conducted on a dataset of over 4 million chess puzzles. Results demonstrate GlickFormer's superior performance compared to the state-of-the-art ChessFormer baseline across multiple metrics. The algorithm's performance has also been recognized through its competitive results in the IEEE BigData 2024 Cup: Predicting Chess Puzzle Difficulty competition. The insights gained from this study have implications for personalized chess training and broader applications in educational technology and cognitive modeling.
Authors: Cheng Lu, Yang Song
Abstract: Consistency models (CMs) are a powerful class of diffusion-based generative models optimized for fast sampling. Most existing CMs are trained using discretized timesteps, which introduce additional hyperparameters and are prone to discretization errors. While continuous-time formulations can mitigate these issues, their success has been limited by training instability. To address this, we propose a simplified theoretical framework that unifies previous parameterizations of diffusion models and CMs, identifying the root causes of instability. Based on this analysis, we introduce key improvements in diffusion process parameterization, network architecture, and training objectives. These changes enable us to train continuous-time CMs at an unprecedented scale, reaching 1.5B parameters on ImageNet 512x512. Our proposed training algorithm, using only two sampling steps, achieves FID scores of 2.06 on CIFAR-10, 1.48 on ImageNet 64x64, and 1.88 on ImageNet 512x512, narrowing the gap in FID scores with the best existing diffusion models to within 10%.
Authors: Alan T. L. Bacellar, Zachary Susskind, Mauricio Breternitz Jr., Eugene John, Lizy K. John, Priscila M. V. Lima, Felipe M. G. Fran\c{c}a
Abstract: We introduce the Differentiable Weightless Neural Network (DWN), a model based on interconnected lookup tables. Training of DWNs is enabled by a novel Extended Finite Difference technique for approximate differentiation of binary values. We propose Learnable Mapping, Learnable Reduction, and Spectral Regularization to further improve the accuracy and efficiency of these models. We evaluate DWNs in three edge computing contexts: (1) an FPGA-based hardware accelerator, where they demonstrate superior latency, throughput, energy efficiency, and model area compared to state-of-the-art solutions, (2) a low-power microcontroller, where they achieve preferable accuracy to XGBoost while subject to stringent memory constraints, and (3) ultra-low-cost chips, where they consistently outperform small models in both accuracy and projected hardware area. DWNs also compare favorably against leading approaches for tabular datasets, with higher average rank. Overall, our work positions DWNs as a pioneering solution for edge-compatible high-throughput neural networks.
Authors: Asher Trockman, Hrayr Harutyunyan, J. Zico Kolter, Sanjiv Kumar, Srinadh Bhojanapalli
Abstract: Recent work has shown that state space models such as Mamba are significantly worse than Transformers on recall-based tasks due to the fact that their state size is constant with respect to their input sequence length. But in practice, state space models have fairly large state sizes, and we conjecture that they should be able to perform much better at these tasks than previously reported. We investigate whether their poor copying and recall performance could be due in part to training difficulties rather than fundamental capacity constraints. Based on observations of their "attention" maps, we propose a structured initialization technique that allows state space layers to more readily mimic attention. Across a variety of architecture settings, our initialization makes it substantially easier for Mamba to learn to copy and do associative recall from scratch.
Authors: Severi Rissanen, Markus Heinonen, Arno Solin
Abstract: The covariance for clean data given a noisy observation is an important quantity in many conditional generation methods for diffusion models. Current methods require heavy test-time computation, altering the standard diffusion training process or denoiser architecture, or making heavy approximations. We propose a new framework that sidesteps these issues by using covariance information that is available for free from training data and the curvature of the generative trajectory, which is linked to the covariance through the second-order Tweedie's formula. We integrate these sources of information using {\em (i)} a novel method to transfer covariance estimates across noise levels and (ii) low-rank updates in a given noise level. We validate the method on linear inverse problems, where it outperforms recent baselines, especially with fewer diffusion steps.
Authors: Zhitong Xu, Da Long, Yiming Xu, Guang Yang, Shandian Zhe, Houman Owhadi
Abstract: This paper introduces a novel kernel learning framework toward efficiently solving nonlinear partial differential equations (PDEs). In contrast to the state-of-the-art kernel solver that embeds differential operators within kernels, posing challenges with a large number of collocation points, our approach eliminates these operators from the kernel. We model the solution using a standard kernel interpolation form and differentiate the interpolant to compute the derivatives. Our framework obviates the need for complex Gram matrix construction between solutions and their derivatives, allowing for a straightforward implementation and scalable computation. As an instance, we allocate the collocation points on a grid and adopt a product kernel, which yields a Kronecker product structure in the interpolation. This structure enables us to avoid computing the full Gram matrix, reducing costs and scaling efficiently to a large number of collocation points. We provide a proof of the convergence and rate analysis of our method under appropriate regularity assumptions. In numerical experiments, we demonstrate the advantages of our method in solving several benchmark PDEs.
Authors: Karen Medlin, Sven Leyffer, Krishnan Raghavan
Abstract: Data rebalancing techniques, including oversampling and undersampling, are a common approach to addressing the challenges of imbalanced data. To tackle unresolved problems related to both oversampling and undersampling, we propose a new undersampling approach that: (i) avoids the pitfalls of noise and overlap caused by synthetic data and (ii) avoids the pitfall of under-fitting caused by random undersampling. Instead of undersampling majority data randomly, our method undersamples datapoints based on their ability to improve model loss. Using improved model loss as a proxy measurement for classification performance, our technique assesses a datapoint's impact on loss and rejects those unable to improve it. In so doing, our approach rejects majority datapoints redundant to datapoints already accepted and, thereby, finds an optimal subset of majority training data for classification. The accept/reject component of our algorithm is motivated by a bilevel optimization problem uniquely formulated to identify the optimal training set we seek. Experimental results show our proposed technique with F1 scores up to 10% higher than state-of-the-art methods.
Authors: Kola Ayonrinde, Michael T. Pearce, Lee Sharkey
Abstract: Sparse Autoencoders (SAEs) have emerged as a useful tool for interpreting the internal representations of neural networks. However, naively optimising SAEs for reconstruction loss and sparsity results in a preference for SAEs that are extremely wide and sparse. We present an information-theoretic framework for interpreting SAEs as lossy compression algorithms for communicating explanations of neural activations. We appeal to the Minimal Description Length (MDL) principle to motivate explanations of activations which are both accurate and concise. We further argue that interpretable SAEs require an additional property, "independent additivity": features should be able to be understood separately. We demonstrate an example of applying our MDL-inspired framework by training SAEs on MNIST handwritten digits and find that SAE features representing significant line segments are optimal, as opposed to SAEs with features for memorised digits from the dataset or small digit fragments. We argue that using MDL rather than sparsity may avoid potential pitfalls with naively maximising sparsity such as undesirable feature splitting and that this framework naturally suggests new hierarchical SAE architectures which provide more concise explanations.
Authors: Jinyu Liu, Hongye Guo, Yun Li, Qinghu Tang, Fuquan Huang, Tunan Chen, Haiwang Zhong, Qixin Chen
Abstract: Over the past decade, bidding in power markets has attracted widespread attention. Reinforcement Learning (RL) has been widely used for power market bidding as a powerful AI tool to make decisions under real-world uncertainties. However, current RL methods mostly employ low dimensional bids, which significantly diverge from the N price-power pairs commonly used in the current power markets. The N-pair bidding format is denoted as High Dimensional Bids (HDBs), which has not been fully integrated into the existing RL-based bidding methods. The loss of flexibility in current RL bidding methods could greatly limit the bidding profits and make it difficult to tackle the rising uncertainties brought by renewable energy generations. In this paper, we intend to propose a framework to fully utilize HDBs for RL-based bidding methods. First, we employ a special type of neural network called Neural Network Supply Functions (NNSFs) to generate HDBs in the form of N price-power pairs. Second, we embed the NNSF into a Markov Decision Process (MDP) to make it compatible with most existing RL methods. Finally, experiments on Energy Storage Systems (ESSs) in the PJM Real-Time (RT) power market show that the proposed bidding method with HDBs can significantly improve bidding flexibility, thereby improving the profit of the state-of-the-art RL bidding methods.
Authors: Hanbo Huang, Yihan Li, Bowen Jiang, Lin Liu, Ruoyu Sun, Zhuotao Liu, Shiyu Liang
Abstract: Closed-source large language models deliver strong performance but have limited downstream customizability. Semi-open models, combining both closed-source and public layers, were introduced to improve customizability. However, parameters in the closed-source layers are found vulnerable to recovery attacks. In this paper, we explore the design of semi-open models with fewer closed-source layers, aiming to increase customizability while ensuring resilience to recovery attacks. We analyze the contribution of closed-source layer to the overall resilience and theoretically prove that in a deep transformer-based model, there exists a transition layer such that even small recovery errors in layers before this layer can lead to recovery failure. Building on this, we propose \textbf{SCARA}, a novel approach that keeps only a few bottom layers as closed-source. SCARA employs a fine-tuning-free metric to estimate the maximum number of layers that can be publicly accessible for customization. We apply it to five models (1.3B to 70B parameters) to construct semi-open models, validating their customizability on six downstream tasks and assessing their resilience against various recovery attacks on sixteen benchmarks. We compare SCARA to baselines and observe that it generally improves downstream customization performance and offers similar resilience with over \textbf{10} times fewer closed-source parameters. We empirically investigate the existence of transition layers, analyze the effectiveness of our scheme and finally discuss its limitations.
Authors: Haiquan Qiu, Shuzhi Liu, Quanming Yao
Abstract: Complex networks describe important structures in nature and society, composed of nodes and the edges that connect them. The evolution of these networks is typically described by dynamics, which are labor-intensive and require expert knowledge to derive. However, because the complex network involves noisy observations from multiple trajectories of nodes, existing symbolic regression methods are either not applicable or ineffective on its dynamics. In this paper, we propose Physically Inspired Neural Dynamics Symbolic Regression (PI-NDSR), a method based on neural networks and genetic programming to automatically learn the symbolic expression of dynamics. Our method consists of two key components: a Physically Inspired Neural Dynamics (PIND) to augment and denoise trajectories through observed trajectory interpolation; and a coordinated genetic search algorithm to derive symbolic expressions. This algorithm leverages references of node dynamics and edge dynamics from neural dynamics to avoid overfitted expressions in symbolic space. We evaluate our method on synthetic datasets generated by various dynamics and real datasets on disease spreading. The results demonstrate that PI-NDSR outperforms the existing method in terms of both recovery probability and error.
Authors: Dongxie Wen, Xiao Zhang, Zhewei Wei
Abstract: Online Kernel Learning (OKL) has attracted considerable research interest due to its promising predictive performance in streaming environments. Second-order approaches are particularly appealing for OKL as they often offer substantial improvements in regret guarantees. However, existing second-order OKL approaches suffer from at least quadratic time complexity with respect to the pre-set budget, rendering them unsuitable for meeting the real-time demands of large-scale streaming recommender systems. The singular value decomposition required to obtain explicit feature mapping is also computationally expensive due to the complete decomposition process. Moreover, the absence of incremental updates to manage approximate kernel space causes these algorithms to perform poorly in adversarial environments and real-world streaming recommendation datasets. To address these issues, we propose FORKS, a fast incremental matrix sketching and decomposition approach tailored for second-order OKL. FORKS constructs an incremental maintenance paradigm for second-order kernelized gradient descent, which includes incremental matrix sketching for kernel approximation and incremental matrix decomposition for explicit feature mapping construction. Theoretical analysis demonstrates that FORKS achieves a logarithmic regret guarantee on par with other second-order approaches while maintaining a linear time complexity w.r.t. the budget, significantly enhancing efficiency over existing approaches. We validate the performance of FORKS through extensive experiments conducted on real-world streaming recommendation datasets, demonstrating its superior scalability and robustness against adversarial attacks.
Authors: Jiajun Zhou, Xuanze Chen, Chenxuan Xie, Yu Shanqing, Qi Xuan, Xiaoniu Yang
Abstract: Graph Transformer (GT), as a special type of Graph Neural Networks (GNNs), utilizes multi-head attention to facilitate high-order message passing. However, this also imposes several limitations in node classification applications: 1) nodes are susceptible to global noise; 2) self-attention computation cannot scale well to large graphs. In this work, we conduct extensive observational experiments to explore the adaptability of the GT architecture in node classification tasks and draw several conclusions: the current multi-head self-attention module in GT can be completely replaceable, while the feed-forward neural network module proves to be valuable. Based on this, we decouple the propagation (P) and transformation (T) of GNNs and explore a powerful GT architecture, named GNNFormer, which is based on the P/T combination message passing and adapted for node classification in both homophilous and heterophilous scenarios. Extensive experiments on 12 benchmark datasets demonstrate that our proposed GT architecture can effectively adapt to node classification tasks without being affected by global noise and computational efficiency limitations.
Authors: Rikuto Kotoge, Zheng Chen, Tasuku Kimura, Yasuko Matsubara, Takufumi Yanagisawa, Haruhiko Kishima, Yasushi Sakurai
Abstract: While end-to-end multi-channel electroencephalography (EEG) learning approaches have shown significant promise, their applicability is often constrained in neurological diagnostics, such as intracranial EEG resources. When provided with a single-channel EEG, how can we learn representations that are robust to multi-channels and scalable across varied tasks, such as seizure prediction? In this paper, we present SplitSEE, a structurally splittable framework designed for effective temporal-frequency representation learning in single-channel EEG. The key concept of SplitSEE is a self-supervised framework incorporating a deep clustering task. Given an EEG, we argue that the time and frequency domains are two distinct perspectives, and hence, learned representations should share the same cluster assignment. To this end, we first propose two domain-specific modules that independently learn domain-specific representation and address the temporal-frequency tradeoff issue in conventional spectrogram-based methods. Then, we introduce a novel clustering loss to measure the information similarity. This encourages representations from both domains to coherently describe the same input by assigning them a consistent cluster. SplitSEE leverages a pre-training-to-fine-tuning framework within a splittable architecture and has following properties: (a) Effectiveness: it learns representations solely from single-channel EEG but has even outperformed multi-channel baselines. (b) Robustness: it shows the capacity to adapt across different channels with low performance variance. Superior performance is also achieved with our collected clinical dataset. (c) Scalability: With just one fine-tuning epoch, SplitSEE achieves high and stable performance using partial model layers.
Authors: Alireza Khodamoradi, Kristof Denolf, Eric Dellinger
Abstract: Quantization reduces the model's hardware costs, such as data movement, storage, and operations like multiply and addition. It also affects the model's behavior by degrading the output quality. Therefore, there is a need for methods that preserve the model's behavior when quantizing model parameters. More exotic numerical encodings, such as block-scaled number formats, have shown advantages for utilizing a fixed bit budget to encode model parameters. This paper presents error diffusion (ED), a hyperparameter-free method for post-training quantization with support for block-scaled data formats. Our approach does not rely on backpropagation or Hessian information. We describe how to improve the quantization process by viewing the neural model as a composite function and diffusing the quantization error in every layer. In addition, we introduce TensorCast, an open-source library based on PyTorch to emulate a variety of number formats, including the block-scaled ones, to aid the research in neural model quantization. We demonstrate the efficacy of our algorithm through rigorous testing on various architectures, including vision and large language models (LLMs), where it consistently delivers competitive results. Our experiments confirm that block-scaled data formats provide a robust choice for post-training quantization and could be used effectively to enhance the practical deployment of advanced neural networks.
Authors: Hassan Ali, Surya Nepal, Salil S. Kanhere, Sanjay Jha
Abstract: Recent works have shown that Federated Learning (FL) is vulnerable to backdoor attacks. Existing defenses cluster submitted updates from clients and select the best cluster for aggregation. However, they often rely on unrealistic assumptions regarding client submissions and sampled clients population while choosing the best cluster. We show that in realistic FL settings, state-of-the-art (SOTA) defenses struggle to perform well against backdoor attacks in FL. To address this, we highlight that backdoored submissions are adversarially biased and overconfident compared to clean submissions. We, therefore, propose an Adversarially Guided Stateful Defense (AGSD) against backdoor attacks on Deep Neural Networks (DNNs) in FL scenarios. AGSD employs adversarial perturbations to a small held-out dataset to compute a novel metric, called the trust index, that guides the cluster selection without relying on any unrealistic assumptions regarding client submissions. Moreover, AGSD maintains a trust state history of each client that adaptively penalizes backdoored clients and rewards clean clients. In realistic FL settings, where SOTA defenses mostly fail to resist attacks, AGSD mostly outperforms all SOTA defenses with minimal drop in clean accuracy (5% in the worst-case compared to best accuracy) even when (a) given a very small held-out dataset -- typically AGSD assumes 50 samples (<= 0.1% of the training data) and (b) no heldout dataset is available, and out-of-distribution data is used instead. For reproducibility, our code will be openly available at: https://github.com/hassanalikhatim/AGSD.
Authors: Jingyang Li, Jiachun Pan, Vincent Y. F. Tan, Kim-Chuan Toh, Pan Zhou
Abstract: Semi-supervised learning (SSL), exemplified by FixMatch (Sohn et al., 2020), has shown significant generalization advantages over supervised learning (SL), particularly in the context of deep neural networks (DNNs). However, it is still unclear, from a theoretical standpoint, why FixMatch-like SSL algorithms generalize better than SL on DNNs. In this work, we present the first theoretical justification for the enhanced test accuracy observed in FixMatch-like SSL applied to DNNs by taking convolutional neural networks (CNNs) on classification tasks as an example. Our theoretical analysis reveals that the semantic feature learning processes in FixMatch and SL are rather different. In particular, FixMatch learns all the discriminative features of each semantic class, while SL only randomly captures a subset of features due to the well-known lottery ticket hypothesis. Furthermore, we show that our analysis framework can be applied to other FixMatch-like SSL methods, e.g., FlexMatch, FreeMatch, Dash, and SoftMatch. Inspired by our theoretical analysis, we develop an improved variant of FixMatch, termed Semantic-Aware FixMatch (SA-FixMatch). Experimental results corroborate our theoretical findings and the enhanced generalization capability of SA-FixMatch.
Authors: Xuyu Zhang, Haofan Huang, Dawei Zhang, Songlin Zhuang, Shensheng Han, Puxiang Lai, Honglin Liu
Abstract: Deep learning has been extensively used in various fields, such as phase imaging, 3D imaging reconstruction, phase unwrapping, and laser speckle reduction, particularly for complex problems that lack analytic models. Its data-driven nature allows for implicit construction of mathematical relationships within the network through training with abundant data. However, a critical challenge in practical applications is the generalization issue, where a network trained on one dataset struggles to recognize an unknown target from a different dataset. In this study, we investigate imaging through scattering media and discover that the mathematical relationship learned by the network is an approximation dependent on the training dataset, rather than the true mapping relationship of the model. We demonstrate that enhancing the diversity of the training dataset can improve this approximation, thereby achieving generalization across different datasets, as the mapping relationship of a linear physical model is independent of inputs. This study elucidates the nature of generalization across different datasets and provides insights into the design of training datasets to ultimately address the generalization issue in various deep learning-based applications.
Authors: Peter Vamplew, Conor F Hayes, Cameron Foale, Richard Dazeley, Hadassah Harland
Abstract: Reinforcement learning (RL) is a valuable tool for the creation of AI systems. However it may be problematic to adequately align RL based on scalar rewards if there are multiple conflicting values or stakeholders to be considered. Over the last decade multi-objective reinforcement learning (MORL) using vector rewards has emerged as an alternative to standard, scalar RL. This paper provides an overview of the role which MORL can play in creating pluralistically-aligned AI.
Authors: Peter Eckmann, Dongxia Wu, Germano Heinzelmann, Michael K Gilson, Rose Yu
Abstract: Current generative models for drug discovery primarily use molecular docking as an oracle to guide the generation of active compounds. However, such models are often not useful in practice because even compounds with high docking scores do not consistently show experimental activity. More accurate methods for activity prediction exist, such as molecular dynamics based binding free energy calculations, but they are too computationally expensive to use in a generative model. To address this challenge, we propose Multi-Fidelity Latent space Active Learning (MF-LAL), a generative modeling framework that integrates a set of oracles with varying cost-accuracy tradeoffs. Unlike previous approaches that separately learn the surrogate model and generative model, MF-LAL combines the generative and multi-fidelity surrogate models into a single framework, allowing for more accurate activity prediction and higher quality samples. We train MF-LAL with a novel active learning algorithm to further reduce computational cost. Our experiments on two disease-relevant proteins show that MF-LAL produces compounds with significantly better binding free energy scores than other single and multi-fidelity approaches.
Authors: Jiayu Chen, Wentse Chen, Jeff Schneider
Abstract: Offline reinforcement learning (RL) is a powerful approach for data-driven decision-making and control. Compared to model-free methods, offline model-based reinforcement learning (MBRL) explicitly learns world models from a static dataset and uses them as surrogate simulators, improving the data efficiency and enabling the learned policy to potentially generalize beyond the dataset support. However, there could be various MDPs that behave identically on the offline dataset and so dealing with the uncertainty about the true MDP can be challenging. In this paper, we propose modeling offline MBRL as a Bayes Adaptive Markov Decision Process (BAMDP), which is a principled framework for addressing model uncertainty. We further introduce a novel Bayes Adaptive Monte-Carlo planning algorithm capable of solving BAMDPs in continuous state and action spaces with stochastic transitions. This planning process is based on Monte Carlo Tree Search and can be integrated into offline MBRL as a policy improvement operator in policy iteration. Our ``RL + Search" framework follows in the footsteps of superhuman AIs like AlphaZero, improving on current offline MBRL methods by incorporating more computation input. The proposed algorithm significantly outperforms state-of-the-art model-based and model-free offline RL methods on twelve D4RL MuJoCo benchmark tasks and three target tracking tasks in a challenging, stochastic tokamak control simulator.
Authors: Naveen Gupta, Medha Sawhney, Arka Daw, Youzuo Lin, Anuj Karpatne
Abstract: In subsurface imaging, learning the mapping from velocity maps to seismic waveforms (forward problem) and waveforms to velocity (inverse problem) is important for several applications. While traditional techniques for solving forward and inverse problems are computationally prohibitive, there is a growing interest in leveraging recent advances in deep learning to learn the mapping between velocity maps and seismic waveform images directly from data. Despite the variety of architectures explored in previous works, several open questions still remain unanswered such as the effect of latent space sizes, the importance of manifold learning, the complexity of translation models, and the value of jointly solving forward and inverse problems. We propose a unified framework to systematically characterize prior research in this area termed the Generalized Forward-Inverse (GFI) framework, building on the assumption of manifolds and latent space translations. We show that GFI encompasses previous works in deep learning for subsurface imaging, which can be viewed as specific instantiations of GFI. We also propose two new model architectures within the framework of GFI: Latent U-Net and Invertible X-Net, leveraging the power of U-Nets for domain translation and the ability of IU-Nets to simultaneously learn forward and inverse translations, respectively. We show that our proposed models achieve state-of-the-art (SOTA) performance for forward and inverse problems on a wide range of synthetic datasets, and also investigate their zero-shot effectiveness on two real-world-like datasets.
Authors: Jiaheng Hu, Zizhao Wang, Peter Stone, Roberto Mart\'in-Mart\'in
Abstract: A hallmark of intelligent agents is the ability to learn reusable skills purely from unsupervised interaction with the environment. However, existing unsupervised skill discovery methods often learn entangled skills where one skill variable simultaneously influences many entities in the environment, making downstream skill chaining extremely challenging. We propose Disentangled Unsupervised Skill Discovery (DUSDi), a method for learning disentangled skills that can be efficiently reused to solve downstream tasks. DUSDi decomposes skills into disentangled components, where each skill component only affects one factor of the state space. Importantly, these skill components can be concurrently composed to generate low-level actions, and efficiently chained to tackle downstream tasks through hierarchical Reinforcement Learning. DUSDi defines a novel mutual-information-based objective to enforce disentanglement between the influences of different skill components, and utilizes value factorization to optimize this objective efficiently. Evaluated in a set of challenging environments, DUSDi successfully learns disentangled skills, and significantly outperforms previous skill discovery methods when it comes to applying the learned skills to solve downstream tasks. Code and skills visualization at jiahenghu.github.io/DUSDi-site/.
Authors: Yingyu Liang, Jiangxuan Long, Zhenmei Shi, Zhao Song, Yufa Zhou
Abstract: Large Language Models (LLMs) have shown immense potential in enhancing various aspects of our daily lives, from conversational AI to search and AI assistants. However, their growing capabilities come at the cost of extremely large model sizes, making deployment on edge devices challenging due to memory and computational constraints. This paper introduces a novel approach to LLM weight pruning that directly optimizes for approximating the attention matrix, a core component of transformer architectures. Unlike existing methods that focus on linear approximations, our approach accounts for the non-linear nature of the Softmax attention mechanism. We provide theoretical guarantees for the convergence of our Gradient Descent-based optimization method to a near-optimal pruning mask solution. Our preliminary empirical results demonstrate the effectiveness of this approach in maintaining model performance while significantly reducing computational costs. This work establishes a new theoretical foundation for pruning algorithm design in LLMs, potentially paving the way for more efficient LLM inference on resource-constrained devices.
Authors: Mahdi Alikhasi, Levi H. S. Lelis
Abstract: In reinforcement learning, agents often learn policies for specific tasks without the ability to generalize this knowledge to related tasks. This paper introduces an algorithm that attempts to address this limitation by decomposing neural networks encoding policies for Markov Decision Processes into reusable sub-policies, which are used to synthesize temporally extended actions, or options. We consider neural networks with piecewise linear activation functions, so that they can be mapped to an equivalent tree that is similar to oblique decision trees. Since each node in such a tree serves as a function of the input of the tree, each sub-tree is a sub-policy of the main policy. We turn each of these sub-policies into options by wrapping it with while-loops of varied number of iterations. Given the large number of options, we propose a selection mechanism based on minimizing the Levin loss for a uniform policy on these options. Empirical results in two grid-world domains where exploration can be difficult confirm that our method can identify useful options, thereby accelerating the learning process on similar but different tasks.
Authors: Xinpeng Wang, Xiaoying Tang
Abstract: Domain Generalization (DG) aims to train models that can effectively generalize to unseen domains. However, in the context of Federated Learning (FL), where clients collaboratively train a model without directly sharing their data, most existing DG algorithms are not directly applicable to the FL setting due to privacy constraints, as well as the limited data quantity and domain diversity at each client. To tackle these challenges, we propose FedCCRL, a novel federated domain generalization method that significantly improves the model's ability to generalize to unseen domains without compromising privacy or incurring excessive computational and communication costs. Specifically, we adapt MixStyle to the federated setting to transfer domain-specific features while AugMix is employed to perturb domain-invariant features. Furthermore, we leverage supervised contrastive loss for representation alignment and utilize Jensen-Shannon divergence to ensure consistent predictions between original and augmented samples. Extensive experimental results demonstrate that FedCCRL achieves the state-of-the-art performances on the PACS, OfficeHome and miniDomainNet datasets across varying numbers of clients. Code is available at https://github.com/SanphouWang/FedCCRL.
Authors: Bo Chen, Xiaoyu Li, Yingyu Liang, Zhenmei Shi, Zhao Song
Abstract: In-context learning has been recognized as a key factor in the success of Large Language Models (LLMs). It refers to the model's ability to learn patterns on the fly from provided in-context examples in the prompt during inference. Previous studies have demonstrated that the Transformer architecture used in LLMs can implement a single-step gradient descent update by processing in-context examples in a single forward pass. Recent work has further shown that, during in-context learning, a looped Transformer can implement multi-step gradient descent updates in forward passes. However, their theoretical results require an exponential number of in-context examples, $n = \exp(\Omega(T))$, where $T$ is the number of loops or passes, to achieve a reasonably low error. In this paper, we study linear looped Transformers in-context learning on linear vector generation tasks. We show that linear looped Transformers can implement multi-step gradient descent efficiently for in-context learning. Our results demonstrate that as long as the input data has a constant condition number, e.g., $n = O(d)$, the linear looped Transformers can achieve a small error by multi-step gradient descent during in-context learning. Furthermore, our preliminary experiments validate our theoretical analysis. Our findings reveal that the Transformer architecture possesses a stronger in-context learning capability than previously understood, offering new insights into the mechanisms behind LLMs and potentially guiding the better design of efficient inference algorithms for LLMs.
Authors: Hung-Chieh Fang, Po-Yi Lu, Hsuan-Tien Lin
Abstract: Universal Domain Adaptation (UniDA) aims to transfer knowledge from a labeled source domain to an unlabeled target domain without assuming how much the label-sets of the two domains intersect. The goal of UniDA is to achieve robust performance on the target domain across different intersection levels. However, existing literature has not sufficiently explored performance under extreme intersection levels. Our experiments reveal that state-of-the-art methods struggle when the source domain has significantly more non-overlapping classes than overlapping ones, a setting we refer to as Extreme UniDA. In this paper, we demonstrate that classical partial domain alignment, which focuses on aligning only overlapping-class data between domains, is limited in mitigating the bias of feature extractors toward source-private classes in extreme UniDA scenarios. We argue that feature extractors trained with source supervised loss distort the intrinsic structure of the target data due to the inherent differences between source-private classes and the target data. To mitigate this bias, we propose using self-supervised learning to preserve the structure of the target data. Our approach can be easily integrated into existing frameworks. We apply the proposed approach to two distinct training paradigms-adversarial-based and optimal-transport-based-and show consistent improvements across various intersection levels, with significant gains in extreme UniDA settings.
Authors: Nicholas M. Boffi, Arthur Jacot, Stephen Tu, Ingvar Ziemann
Abstract: Diffusion-based generative models provide a powerful framework for learning to sample from a complex target distribution. The remarkable empirical success of these models applied to high-dimensional signals, including images and video, stands in stark contrast to classical results highlighting the curse of dimensionality for distribution recovery. In this work, we take a step towards understanding this gap through a careful analysis of learning diffusion models over the Barron space of single layer neural networks. In particular, we show that these shallow models provably adapt to simple forms of low dimensional structure, thereby avoiding the curse of dimensionality. We combine our results with recent analyses of sampling with diffusion models to provide an end-to-end sample complexity bound for learning to sample from structured distributions. Importantly, our results do not require specialized architectures tailored to particular latent structures, and instead rely on the low-index structure of the Barron space to adapt to the underlying distribution.
Authors: Abhijit Manatkar, Devarsh Patel, Hima Patel, Naresh Manwani
Abstract: Automating end-to-end Exploratory Data Analysis (AutoEDA) is a challenging open problem, often tackled through Reinforcement Learning (RL) by learning to predict a sequence of analysis operations (FILTER, GROUP, etc). Defining rewards for each operation is a challenging task and existing methods rely on various \emph{interestingness measures} to craft reward functions to capture the importance of each operation. In this work, we argue that not all of the essential features of what makes an operation important can be accurately captured mathematically using rewards. We propose an AutoEDA model trained through imitation learning from expert EDA sessions, bypassing the need for manually defined interestingness measures. Our method, based on generative adversarial imitation learning (GAIL), generalizes well across datasets, even with limited expert data. We also introduce a novel approach for generating synthetic EDA demonstrations for training. Our method outperforms the existing state-of-the-art end-to-end EDA approach on benchmarks by upto 3x, showing strong performance and generalization, while naturally capturing diverse interestingness measures in generated EDA sessions.
Authors: Li Wu, Wenbin Pei, Jiulong Jiao, Qiang Zhang
Abstract: Multivariate Time series forecasting is crucial in domains such as transportation, meteorology, and finance, especially for predicting extreme weather events. State-of-the-art methods predominantly rely on Transformer architectures, which utilize attention mechanisms to capture temporal dependencies. However, these methods are hindered by quadratic time complexity, limiting the model's scalability with respect to input sequence length. This significantly restricts their practicality in the real world. Mamba, based on state space models (SSM), provides a solution with linear time complexity, increasing the potential for efficient forecasting of sequential data. In this study, we propose UmambaTSF, a novel long-term time series forecasting framework that integrates multi-scale feature extraction capabilities of U-shaped encoder-decoder multilayer perceptrons (MLP) with Mamba's long sequence representation. To improve performance and efficiency, the Mamba blocks introduced in the framework adopt a refined residual structure and adaptable design, enabling the capture of unique temporal signals and flexible channel processing. In the experiments, UmambaTSF achieves state-of-the-art performance and excellent generality on widely used benchmark datasets while maintaining linear time complexity and low memory consumption.
Authors: Yekun Ke, Xiaoyu Li, Yingyu Liang, Zhenmei Shi, Zhao Song
Abstract: Recent empirical studies have identified fixed point iteration phenomena in deep neural networks, where the hidden state tends to stabilize after several layers, showing minimal change in subsequent layers. This observation has spurred the development of practical methodologies, such as accelerating inference by bypassing certain layers once the hidden state stabilizes, selectively fine-tuning layers to modify the iteration process, and implementing loops of specific layers to maintain fixed point iterations. Despite these advancements, the understanding of fixed point iterations remains superficial, particularly in high-dimensional spaces, due to the inadequacy of current analytical tools. In this study, we conduct a detailed analysis of fixed point iterations in a vector-valued function modeled by neural networks. We establish a sufficient condition for the existence of multiple fixed points of looped neural networks based on varying input regions. Additionally, we expand our examination to include a robust version of fixed point iterations. To demonstrate the effectiveness and insights provided by our approach, we provide case studies that looped neural networks may exist $2^d$ number of robust fixed points under exponentiation or polynomial activation functions, where $d$ is the feature dimension. Furthermore, our preliminary empirical results support our theoretical findings. Our methodology enriches the toolkit available for analyzing fixed point iterations of deep neural networks and may enhance our comprehension of neural network mechanisms.
Authors: Pankayaraj Pathmanathan, Udari Madhushani Sehwag, Michael-Andrei Panaitescu-Liess, Furong Huang
Abstract: With the growing adoption of reinforcement learning with human feedback (RLHF) for aligning large language models (LLMs), the risk of backdoor installation during alignment has increased, leading to unintended and harmful behaviors. Existing backdoor triggers are typically limited to fixed word patterns, making them detectable during data cleaning and easily removable post-poisoning. In this work, we explore the use of prompt-specific paraphrases as backdoor triggers, enhancing their stealth and resistance to removal during LLM alignment. We propose AdvBDGen, an adversarially fortified generative fine-tuning framework that automatically generates prompt-specific backdoors that are effective, stealthy, and transferable across models. AdvBDGen employs a generator-discriminator pair, fortified by an adversary, to ensure the installability and stealthiness of backdoors. It enables the crafting and successful installation of complex triggers using as little as 3% of the fine-tuning data. Once installed, these backdoors can jailbreak LLMs during inference, demonstrate improved stability against perturbations compared to traditional constant triggers, and are more challenging to remove. These findings underscore an urgent need for the research community to develop more robust defenses against adversarial backdoor threats in LLM alignment.
Authors: Yutong He, Pengrui Li, Yipeng Hu, Chuyan Chen, Kun Yuan
Abstract: Subspace optimization algorithms, with GaLore (Zhao et al., 2024) as a representative method, have gained popularity for pre-training or fine-tuning large language models (LLMs) due to their memory efficiency. However, their convergence guarantees remain unclear, particularly in stochastic settings. In this paper, we unexpectedly discover that GaLore does not always converge to the optimal solution and substantiate this finding with an explicit counterexample. We then investigate the conditions under which GaLore can achieve convergence, demonstrating that it does so either in deterministic scenarios or when using a sufficiently large mini-batch size. More significantly, we introduce GoLore (Gradient random Low-rank projection), a novel variant of GaLore that provably converges in stochastic settings, even with standard batch sizes. Our convergence analysis can be readily extended to other sparse subspace optimization algorithms. Finally, we conduct numerical experiments to validate our theoretical results and empirically explore the proposed mechanisms. Codes are available at https://github.com/pkumelon/Golore.
Authors: Jirui Yang, Peng Chen, Zhihui Lu, Ruijun Deng, Qiang Duan, Jianping Zeng
Abstract: Federated Graph Neural Network (FedGNN) is a privacy-preserving machine learning technology that combines federated learning (FL) and graph neural networks (GNNs). It offers a privacy-preserving solution for training GNNs using isolated graph data. Vertical Federated Graph Neural Network (VFGNN) is an important branch of FedGNN, where data features and labels are distributed among participants, and each participant has the same sample space. Due to the difficulty of accessing and modifying distributed data and labels, the vulnerability of VFGNN to backdoor attacks remains largely unexplored. In this context, we propose BVG, the first method for backdoor attacks in VFGNN. Without accessing or modifying labels, BVG uses multi-hop triggers and requires only four target class nodes for an effective backdoor attack. Experiments show that BVG achieves high attack success rates (ASR) across three datasets and three different GNN models, with minimal impact on main task accuracy (MTA). We also evaluate several defense methods, further validating the robustness and effectiveness of BVG. This finding also highlights the need for advanced defense mechanisms to counter sophisticated backdoor attacks in practical VFGNN applications.
Authors: Jinjae Kim, Minjeong Ma, Eunjee Choi, Keunhee Cho, Chanwoo Lee
Abstract: This paper presents a novel approach that leverages Transformer-based multivariate time series model and Machine Learning Ensembles to predict the quality of human sleep, emotional states, and stress levels. A formula to calculate the labels was developed, and the various models were applied to user data. Time Series Transformer was used for labels where time series characteristics are crucial, while Machine Learning Ensembles were employed for labels requiring comprehensive daily activity statistics. Time Series Transformer excels in capturing the characteristics of time series through pre-training, while Machine Learning Ensembles select machine learning models that meet our categorization criteria. The proposed model, TraM, scored 6.10 out of 10 in experiments, demonstrating superior performance compared to other methodologies. The code and configuration for the TraM framework are available at: https://github.com/jin-jae/ETRI-Paper-Contest.
Authors: Zifan Liu, Amin Karbasi, Theodoros Rekatsinas
Abstract: Finetuning foundation models for specific tasks is an emerging paradigm in modern machine learning. The efficacy of task-specific finetuning largely depends on the selection of appropriate training data. We present a framework to select data for task-specific model finetuning, guided by a small but representative set of examples from the target task. To do so, we formulate data selection for task-specific finetuning as an optimization problem with a distribution alignment loss based on optimal transport to capture the discrepancy between the selected data and the target distribution. In addition, we add a regularizer to encourage the diversity of the selected data and incorporate kernel density estimation into the regularizer to reduce the negative effects of near-duplicates among the candidate data. We connect our optimization problem to nearest neighbor search and design efficient algorithms to compute the optimal solution based on approximate nearest neighbor search techniques. We evaluate our method on data selection for both continued pretraining and instruction tuning of language models. We show that instruction tuning using data selected by our method with a 1% selection ratio often outperforms using the full dataset and beats the baseline selection methods by 1.5 points in F1 score on average.
Authors: Juntao Zhao, Wenhao Lu, Sheng Wang, Lingpeng Kong, Chuan Wu
Abstract: Quantization has been substantially adopted to accelerate inference and reduce memory consumption of large language models (LLMs). While activation-weight joint quantization speeds up the inference process through low-precision kernels, we demonstrate that it suffers severe performance degradation on multi-step reasoning tasks, rendering it ineffective. We propose a novel quantization paradigm called QSPEC, which seamlessly integrates two complementary quantization schemes for speculative decoding. Leveraging nearly cost-free execution switching, QSPEC drafts tokens with low-precision, fast activation-weight quantization, and verifies them with high-precision weight-only quantization, effectively combining the strengths of both quantization schemes. Compared to high-precision quantization methods, QSPEC empirically boosts token generation throughput by up to 1.80x without any quality compromise, distinguishing it from other low-precision quantization approaches. This enhancement is also consistent across various serving tasks, model sizes, quantization methods, and batch sizes. Unlike existing speculative decoding techniques, our approach reuses weights and the KV cache, avoiding additional memory overhead. Furthermore, QSPEC offers a plug-and-play advantage without requiring any training. We believe that QSPEC demonstrates unique strengths for future deployment of high-fidelity quantization schemes, particularly in memory-constrained scenarios (e.g., edge devices).
Authors: Yuntian Gu, Xuzheng Chen
Abstract: Multilevel optimization has gained renewed interest in machine learning due to its promise in applications such as hyperparameter tuning and continual learning. However, existing methods struggle with the inherent difficulty of efficiently handling the nested structure. This paper introduces a novel gradient-based approach for multilevel optimization that overcomes these limitations by leveraging a hierarchically structured decomposition of the full gradient and employing advanced propagation techniques. Extending to n-level scenarios, our method significantly reduces computational complexity while improving both solution accuracy and convergence speed. We demonstrate the effectiveness of our approach through numerical experiments, comparing it with existing methods across several benchmarks. The results show a notable improvement in solution accuracy. To the best of our knowledge, this is one of the first algorithms to provide a general version of implicit differentiation with both theoretical guarantees and superior empirical performance.
Authors: Qizhang Li, Xiaochen Yang, Wangmeng Zuo, Yiwen Guo
Abstract: Automatic adversarial prompt generation provides remarkable success in jailbreaking safely-aligned large language models (LLMs). Existing gradient-based attacks, while demonstrating outstanding performance in jailbreaking white-box LLMs, often generate garbled adversarial prompts with chaotic appearance. These adversarial prompts are difficult to transfer to other LLMs, hindering their performance in attacking unknown victim models. In this paper, for the first time, we delve into the semantic meaning embedded in garbled adversarial prompts and propose a novel method that "translates" them into coherent and human-readable natural language adversarial prompts. In this way, we can effectively uncover the semantic information that triggers vulnerabilities of the model and unambiguously transfer it to the victim model, without overlooking the adversarial information hidden in the garbled text, to enhance jailbreak attacks. It also offers a new approach to discovering effective designs for jailbreak prompts, advancing the understanding of jailbreak attacks. Experimental results demonstrate that our method significantly improves the success rate of jailbreak attacks against various safety-aligned LLMs and outperforms state-of-the-arts by large margins. With at most 10 queries, our method achieves an average attack success rate of 81.8% in attacking 7 commercial closed-source LLMs, including GPT and Claude-3 series, on HarmBench. Our method also achieves over 90% attack success rates against Llama-2-Chat models on AdvBench, despite their outstanding resistance to jailbreak attacks. Code at: https://github.com/qizhangli/Adversarial-Prompt-Translator.
URLs: https://github.com/qizhangli/Adversarial-Prompt-Translator.
Authors: Longlong Li, Yipeng Zhang, Guanghui Wang, Kelin Xia
Abstract: Molecular property prediction is a crucial task in the process of Artificial Intelligence-Driven Drug Discovery (AIDD). The challenge of developing models that surpass traditional non-neural network methods continues to be a vibrant area of research. This paper presents a novel graph neural network model-the Kolmogorov-Arnold Network (KAN)-based Graph Neural Network (KA-GNN), which incorporates Fourier series, specifically designed for molecular property prediction. This model maintains the high interpretability characteristic of KAN methods while being extremely efficient in computational resource usage, making it an ideal choice for deployment in resource-constrained environments. Tested and validated on seven public datasets, KA-GNN has shown significant improvements in property predictions over the existing state-of-the-art (SOTA) benchmarks.
Authors: Mathurin Videau (TAU), Mariia Zameshina (LIGM), Alessandro Leite (TAU), Laurent Najman (LIGM), Marc Schoenauer (TAU), Olivier Teytaud (TAU)
Abstract: AfterLearnER (After Learning Evolutionary Retrofitting) consists in applying non-differentiable optimization, including evolutionary methods, to refine fully-trained machine learning models by optimizing a set of carefully chosen parameters or hyperparameters of the model, with respect to some actual, exact, and hence possibly non-differentiable error signal, performed on a subset of the standard validation set. The efficiency of AfterLearnER is demonstrated by tackling non-differentiable signals such as threshold-based criteria in depth sensing, the word error rate in speech re-synthesis, image quality in 3D generative adversarial networks (GANs), image generation via Latent Diffusion Models (LDM), the number of kills per life at Doom, computational accuracy or BLEU in code translation, and human appreciations in image synthesis. In some cases, this retrofitting is performed dynamically at inference time by taking into account user inputs. The advantages of AfterLearnER are its versatility (no gradient is needed), the possibility to use non-differentiable feedback including human evaluations, the limited overfitting, supported by a theoretical study and its anytime behavior. Last but not least, AfterLearnER requires only a minimal amount of feedback, i.e., a few dozens to a few hundreds of scalars, rather than the tens of thousands needed in most related published works. Compared to fine-tuning (typically using the same loss, and gradient-based optimization on a smaller but still big dataset at a fine grain), AfterLearnER uses a minimum amount of data on the real objective function without requiring differentiability.
Authors: Jaehyun Park, Yunho Kim, Sejin Kim, Byung-Jun Lee, Sundong Kim
Abstract: We propose a novel offline reinforcement learning (offline RL) approach, introducing the Diffusion-model-guided Implicit Q-learning with Adaptive Revaluation (DIAR) framework. We address two key challenges in offline RL: out-of-distribution samples and long-horizon problems. We leverage diffusion models to learn state-action sequence distributions and incorporate value functions for more balanced and adaptive decision-making. DIAR introduces an Adaptive Revaluation mechanism that dynamically adjusts decision lengths by comparing current and future state values, enabling flexible long-term decision-making. Furthermore, we address Q-value overestimation by combining Q-network learning with a value function guided by a diffusion model. The diffusion model generates diverse latent trajectories, enhancing policy robustness and generalization. As demonstrated in tasks like Maze2D, AntMaze, and Kitchen, DIAR consistently outperforms state-of-the-art algorithms in long-horizon, sparse-reward environments.
Authors: Dongjoon Lee, Hyeryn Park, Changhee Lee
Abstract: Previous deep learning approaches for survival analysis have primarily relied on ranking losses to improve discrimination performance, which often comes at the expense of calibration performance. To address such an issue, we propose a novel contrastive learning approach specifically designed to enhance discrimination \textit{without} sacrificing calibration. Our method employs weighted sampling within a contrastive learning framework, assigning lower penalties to samples with similar survival outcomes. This aligns well with the assumption that patients with similar event times share similar clinical statuses. Consequently, when augmented with the commonly used negative log-likelihood loss, our approach significantly improves discrimination performance without directly manipulating the model outputs, thereby achieving better calibration. Experiments on multiple real-world clinical datasets demonstrate that our method outperforms state-of-the-art deep survival models in both discrimination and calibration. Through comprehensive ablation studies, we further validate the effectiveness of our approach through quantitative and qualitative analyses.
Authors: Minoo Jafarlou, Mario M. Kubek
Abstract: Labeling datasets is a noteworthy challenge in machine learning, both in terms of cost and time. This research, however, leverages an efficient answer. By exploring label propagation in semi-supervised learning, we can significantly reduce the number of labels required compared to traditional methods. We employ a transductive label propagation method based on the manifold assumption for text classification. Our approach utilizes a graph-based method to generate pseudo-labels for unlabeled data for the text classification task, which are then used to train deep neural networks. By extending labels based on cosine proximity within a nearest neighbor graph from network embeddings, we combine unlabeled data into supervised learning, thereby reducing labeling costs. Based on previous successes in other domains, this study builds and evaluates this approach's effectiveness in sentiment analysis, presenting insights into semi-supervised learning.
Authors: Eric Hanchen Jiang, Zhi Zhang, Dinghuai Zhang, Andrew Lizarraga, Chenheng Xu, Yasi Zhang, Siyan Zhao, Zhengjie Xu, Peiyu Yu, Yuer Tang, Deqian Kong, Ying Nian Wu
Abstract: Advancements in reinforcement learning have led to the development of sophisticated models capable of learning complex decision-making tasks. However, efficiently integrating world models with decision transformers remains a challenge. In this paper, we introduce a novel approach that combines the Dreamer algorithm's ability to generate anticipatory trajectories with the adaptive learning strengths of the Online Decision Transformer. Our methodology enables parallel training where Dreamer-produced trajectories enhance the contextual decision-making of the transformer, creating a bidirectional enhancement loop. We empirically demonstrate the efficacy of our approach on a suite of challenging benchmarks, achieving notable improvements in sample efficiency and reward maximization over existing methods. Our results indicate that the proposed integrated framework not only accelerates learning but also showcases robustness in diverse and dynamic scenarios, marking a significant step forward in model-based reinforcement learning.
Authors: Guanhua Ye, Jifeng He, Weiqing Wang, Zhe Xue, Feifei Kou, Yawen Li
Abstract: Decentralized learning has become crucial for collaborative model training in environments where data privacy and trust are paramount. In web-based applications, clients are liberated from traditional fixed network topologies, enabling the establishment of arbitrary peer-to-peer (P2P) connections. While this flexibility is highly promising, it introduces a fundamental challenge: the optimal selection of neighbors to ensure effective collaboration. To address this, we introduce WPFed, a fully decentralized, web-based learning framework designed to enable globally optimal neighbor selection. WPFed employs a dynamic communication graph and a weighted neighbor selection mechanism. By assessing inter-client similarity through Locality-Sensitive Hashing (LSH) and evaluating model quality based on peer rankings, WPFed enables clients to identify personalized optimal neighbors on a global scale while preserving data privacy. To enhance security and deter malicious behavior, WPFed integrates verification mechanisms for both LSH codes and performance rankings, leveraging blockchain-driven announcements to ensure transparency and verifiability. Through extensive experiments on multiple real-world datasets, we demonstrate that WPFed significantly improves learning outcomes and system robustness compared to traditional federated learning methods. Our findings highlight WPFed's potential to facilitate effective and secure decentralized collaborative learning across diverse and interconnected web environments.
Authors: Seongho Kim, Jihyun Moon, Juntaek Oh, Insu Choi, Joon-Sung Yang
Abstract: The advent of the Attention mechanism and Transformer architecture enables contextually natural text generation and compresses the burden of processing entire source information into singular vectors. Based on these two main ideas, model sizes gradually increases to accommodate more precise and comprehensive information, leading to the current state-of-the-art LLMs being very large, with parameters around 70 billion. As the model sizes are growing, the demand for substantial storage and computational capacity increases. This leads to the development of high-bandwidth memory and accelerators, as well as a variety of model architectures designed to meet these requirements. We note that LLM architectures have increasingly converged. This paper analyzes how these converged architectures perform in terms of layer configurations, operational mechanisms, and model sizes, considering various hyperparameter settings. In this paper, we conduct a concise survey of the history of LLMs by tracing the evolution of their operational improvements. Furthermore, we summarize the performance trends of LLMs under various hyperparameter settings using the RTX 6000, which features the state-of-the-art Ada Lovelace architecture. We conclude that even the same model can exhibit different behaviors depending on the hyperparameters or whether it is deployed in server or edge environments.
Authors: Xihang Yue, Linchao Zhu, Yi Yang
Abstract: Two typical neural models have been extensively studied for operator learning, learning in spatial space via attention mechanism or learning in spectral space via spectral analysis technique such as Fourier Transform. Spatial learning enables point-level flexibility but lacks global continuity constraint, while spectral learning enforces spectral continuity prior but lacks point-wise adaptivity. This work innovatively combines the continuity prior and the point-level flexibility, with the introduced Point-Calibrated Spectral Transform. It achieves this by calibrating the preset spectral eigenfunctions with the predicted point-wise frequency preference via neural gate mechanism. Beyond this, we introduce Point-Calibrated Spectral Neural Operators, which learn operator mappings by approximating functions with the point-level adaptive spectral basis, thereby not only preserving the benefits of spectral prior but also boasting the superior adaptability comparable to the attention mechanism. Comprehensive experiments demonstrate its consistent performance enhancement in extensive PDE solving scenarios.
Authors: Xinting Liao, Weiming Liu, Pengyang Zhou, Fengyuan Yu, Jiahe Xu, Jun Wang, Wenjie Wang, Chaochao Chen, Xiaolin Zheng
Abstract: Federated learning (FL) is a promising machine learning paradigm that collaborates with client models to capture global knowledge. However, deploying FL models in real-world scenarios remains unreliable due to the coexistence of in-distribution data and unexpected out-of-distribution (OOD) data, such as covariate-shift and semantic-shift data. Current FL researches typically address either covariate-shift data through OOD generalization or semantic-shift data via OOD detection, overlooking the simultaneous occurrence of various OOD shifts. In this work, we propose FOOGD, a method that estimates the probability density of each client and obtains reliable global distribution as guidance for the subsequent FL process. Firstly, SM3D in FOOGD estimates score model for arbitrary distributions without prior constraints, and detects semantic-shift data powerfully. Then SAG in FOOGD provides invariant yet diverse knowledge for both local covariate-shift generalization and client performance generalization. In empirical validations, FOOGD significantly enjoys three main advantages: (1) reliably estimating non-normalized decentralized distributions, (2) detecting semantic shift data via score values, and (3) generalizing to covariate-shift data by regularizing feature extractor. The prejoct is open in https://github.com/XeniaLLL/FOOGD-main.git.
Authors: Yuta Oshima, Masahiro Suzuki, Yutaka Matsuo
Abstract: Multimodal variational autoencoders (VAEs) aim to capture shared latent representations by integrating information from different data modalities. A significant challenge is accurately inferring representations from any subset of modalities without training an impractical number (2^M) of inference networks for all possible modality combinations. Mixture-based models simplify this by requiring only as many inference models as there are modalities, aggregating unimodal inferences. However, they suffer from information loss when modalities are missing. Alignment-based VAEs address this by aligning unimodal inference models with a multimodal model through minimizing the Kullback-Leibler (KL) divergence but face issues due to amortization gaps, which compromise inference accuracy. To tackle these problems, we introduce multimodal iterative amortized inference, an iterative refinement mechanism within the multimodal VAE framework. This method overcomes information loss from missing modalities and minimizes the amortization gap by iteratively refining the multimodal inference using all available modalities. By aligning unimodal inference to this refined multimodal posterior, we achieve unimodal inferences that effectively incorporate multimodal information while requiring only unimodal inputs during inference. Experiments on benchmark datasets show that our approach improves inference performance, evidenced by higher linear classification accuracy and competitive cosine similarity, and enhances cross-modal generation, indicated by lower FID scores. This demonstrates that our method enhances inferred representations from unimodal inputs.
Authors: Jaron Maene, Vincent Derkinderen, Pedro Zuidberg Dos Martires
Abstract: A popular approach to neurosymbolic AI involves mapping logic formulas to arithmetic circuits (computation graphs consisting of sums and products) and passing the outputs of a neural network through these circuits. This approach enforces symbolic constraints onto a neural network in a principled and end-to-end differentiable way. Unfortunately, arithmetic circuits are challenging to run on modern AI accelerators as they exhibit a high degree of irregular sparsity. To address this limitation, we introduce knowledge layers (KLay), a new data structure to represent arithmetic circuits that can be efficiently parallelized on GPUs. Moreover, we contribute two algorithms used in the translation of traditional circuit representations to KLay and a further algorithm that exploits parallelization opportunities during circuit evaluations. We empirically show that KLay achieves speedups of multiple orders of magnitude over the state of the art, thereby paving the way towards scaling neurosymbolic AI to larger real-world applications.
Authors: Christopher Iliffe Sprague, Arne Elofsson, Hossein Azizpour
Abstract: Modeling complex systems that evolve toward equilibrium distributions is important in various physical applications, including molecular dynamics and robotic control. These systems often follow the stochastic gradient descent of an underlying energy function, converging to stationary distributions around energy minima. The local covariance of these distributions is shaped by the energy landscape's curvature, often resulting in anisotropic characteristics. While flow-based generative models have gained traction in generating samples from equilibrium distributions in such applications, they predominately employ isotropic conditional probability paths, limiting their ability to capture such covariance structures. In this paper, we introduce Hessian-Informed Flow Matching (HI-FM), a novel approach that integrates the Hessian of an energy function into conditional flows within the flow matching framework. This integration allows HI-FM to account for local curvature and anisotropic covariance structures. Our approach leverages the linearization theorem from dynamical systems and incorporates additional considerations such as time transformations and equivariance. Empirical evaluations on the MNIST and Lennard-Jones particles datasets demonstrate that HI-FM improves the likelihood of test samples.
Authors: Jiacheng Cen, Anyi Li, Ning Lin, Yuxiang Ren, Zihe Wang, Wenbing Huang
Abstract: Equivariant Graph Neural Networks (GNNs) that incorporate E(3) symmetry have achieved significant success in various scientific applications. As one of the most successful models, EGNN leverages a simple scalarization technique to perform equivariant message passing over only Cartesian vectors (i.e., 1st-degree steerable vectors), enjoying greater efficiency and efficacy compared to equivariant GNNs using higher-degree steerable vectors. This success suggests that higher-degree representations might be unnecessary. In this paper, we disprove this hypothesis by exploring the expressivity of equivariant GNNs on symmetric structures, including $k$-fold rotations and regular polyhedra. We theoretically demonstrate that equivariant GNNs will always degenerate to a zero function if the degree of the output representations is fixed to 1 or other specific values. Based on this theoretical insight, we propose HEGNN, a high-degree version of EGNN to increase the expressivity by incorporating high-degree steerable vectors while maintaining EGNN's efficiency through the scalarization trick. Our extensive experiments demonstrate that HEGNN not only aligns with our theoretical analyses on toy datasets consisting of symmetric structures, but also shows substantial improvements on more complicated datasets such as $N$-body and MD17. Our theoretical findings and empirical results potentially open up new possibilities for the research of equivariant GNNs.
Authors: Shi Fu, Yuzhu Chen, Yingjie Wang, Dacheng Tao
Abstract: Understanding the inner mechanisms of black-box foundation models (FMs) is essential yet challenging in artificial intelligence and its applications. Over the last decade, the long-running focus has been on their explainability, leading to the development of post-hoc explainable methods to rationalize the specific decisions already made by black-box FMs. However, these explainable methods have certain limitations in terms of faithfulness, detail capture and resource requirement. Consequently, in response to these issues, a new class of interpretable methods should be considered to unveil the underlying mechanisms in an accurate, comprehensive, heuristic and resource-light way. This survey aims to review interpretable methods that comply with the aforementioned principles and have been successfully applied to FMs. These methods are deeply rooted in machine learning theory, covering the analysis of generalization performance, expressive capability, and dynamic behavior. They provide a thorough interpretation of the entire workflow of FMs, ranging from the inference capability and training dynamics to their ethical implications. Ultimately, drawing upon these interpretations, this review identifies the next frontier research directions for FMs.
Authors: Zhi Wang, Li Zhang, Wenhao Wu, Yuanheng Zhu, Dongbin Zhao, Chunlin Chen
Abstract: A longstanding goal of artificial general intelligence is highly capable generalists that can learn from diverse experiences and generalize to unseen tasks. The language and vision communities have seen remarkable progress toward this trend by scaling up transformer-based models trained on massive datasets, while reinforcement learning (RL) agents still suffer from poor generalization capacity under such paradigms. To tackle this challenge, we propose Meta Decision Transformer (Meta-DT), which leverages the sequential modeling ability of the transformer architecture and robust task representation learning via world model disentanglement to achieve efficient generalization in offline meta-RL. We pretrain a context-aware world model to learn a compact task representation, and inject it as a contextual condition to the causal transformer to guide task-oriented sequence generation. Then, we subtly utilize history trajectories generated by the meta-policy as a self-guided prompt to exploit the architectural inductive bias. We select the trajectory segment that yields the largest prediction error on the pretrained world model to construct the prompt, aiming to encode task-specific information complementary to the world model maximally. Notably, the proposed framework eliminates the requirement of any expert demonstration or domain knowledge at test time. Experimental results on MuJoCo and Meta-World benchmarks across various dataset types show that Meta-DT exhibits superior few and zero-shot generalization capacity compared to strong baselines while being more practical with fewer prerequisites. Our code is available at https://github.com/NJU-RL/Meta-DT.
Authors: Lincen Yang, Matthijs van Leeuwen
Abstract: Conditional density estimation (CDE) goes beyond regression by modeling the full conditional distribution, providing a richer understanding of the data than just the conditional mean in regression. This makes CDE particularly useful in critical application domains. However, interpretable CDE methods are understudied. Current methods typically employ kernel-based approaches, using kernel functions directly for kernel density estimation or as basis functions in linear models. In contrast, despite their conceptual simplicity and visualization suitability, tree-based methods -- which are arguably more comprehensible -- have been largely overlooked for CDE tasks. Thus, we propose the Conditional Density Tree (CDTree), a fully non-parametric model consisting of a decision tree in which each leaf is formed by a histogram model. Specifically, we formalize the problem of learning a CDTree using the minimum description length (MDL) principle, which eliminates the need for tuning the hyperparameter for regularization. Next, we propose an iterative algorithm that, although greedily, searches the optimal histogram for every possible node split. Our experiments demonstrate that, in comparison to existing interpretable CDE methods, CDTrees are both more accurate (as measured by the log-loss) and more robust against irrelevant features. Further, our approach leads to smaller tree sizes than existing tree-based models, which benefits interpretability.
Authors: Viktoria Schuster
Abstract: Sparse autoencoders (SAEs) have lately been used to uncover interpretable latent features in large language models. Here, we explore their potential for decomposing latent representations in complex and high-dimensional biological data, where the underlying variables are often unknown. On simulated data we show that generative hidden variables can be captured in learned representations in the form of superpositions. The degree to which they are learned depends on the completeness of the representations. Superpositions, however, are not identifiable if these generative variables are unknown. SAEs can to some extent recover these variables, yielding interpretable features. Applied to single-cell multi-omics data, we show that an SAE can uncover key biological processes such as carbon dioxide transport and ion homeostasis, which are crucial for red blood cell differentiation and immune function. Our findings highlight how SAEs can be used in advancing interpretability in biological and other scientific domains.
Authors: Mingze Wang, Ruoxi Yu, Weinan E, Lei Wu
Abstract: Transformers have demonstrated exceptional in-context learning capabilities, yet the theoretical understanding of the underlying mechanisms remain limited. A recent work (Elhage et al., 2021) identified a "rich" in-context mechanism known as induction head, contrasting with "lazy" $n$-gram models that overlook long-range dependencies. In this work, we provide both approximation and optimization analyses of how transformers implement induction heads. In the approximation analysis, we formalize both standard and generalized induction head mechanisms, and examine how transformers can efficiently implement them, with an emphasis on the distinct role of each transformer submodule. For the optimization analysis, we study the training dynamics on a synthetic mixed target, composed of a 4-gram and an in-context 2-gram component. This setting enables us to precisely characterize the entire training process and uncover an {\em abrupt transition} from lazy (4-gram) to rich (induction head) mechanisms as training progresses.
Authors: Razmik Arman Khosrovian, Takaharu Yaguchi, Hiroaki Yoshimura, Takashi Matsubara
Abstract: Deep learning has achieved great success in modeling dynamical systems, providing data-driven simulators to predict complex phenomena, even without known governing equations. However, existing models have two major limitations: their narrow focus on mechanical systems and their tendency to treat systems as monolithic. These limitations reduce their applicability to dynamical systems in other domains, such as electrical and hydraulic systems, and to coupled systems. To address these limitations, we propose Poisson-Dirac Neural Networks (PoDiNNs), a novel framework based on the Dirac structure that unifies the port-Hamiltonian and Poisson formulations from geometric mechanics. This framework enables a unified representation of various dynamical systems across multiple domains as well as their interactions and degeneracies arising from couplings. Our experiments demonstrate that PoDiNNs offer improved accuracy and interpretability in modeling unknown coupled dynamical systems from data.
Authors: Chengting Yu, Lei Liu, Gaoang Wang, Erping Li, Aili Wang
Abstract: Recent insights have revealed that rate-coding is a primary form of information representation captured by surrogate-gradient-based Backpropagation Through Time (BPTT) in training deep Spiking Neural Networks (SNNs). Motivated by these findings, we propose rate-based backpropagation, a training strategy specifically designed to exploit rate-based representations to reduce the complexity of BPTT. Our method minimizes reliance on detailed temporal derivatives by focusing on averaged dynamics, streamlining the computational graph to reduce memory and computational demands of SNNs training. We substantiate the rationality of the gradient approximation between BPTT and the proposed method through both theoretical analysis and empirical observations. Comprehensive experiments on CIFAR-10, CIFAR-100, ImageNet, and CIFAR10-DVS validate that our method achieves comparable performance to BPTT counterparts, and surpasses state-of-the-art efficient training techniques. By leveraging the inherent benefits of rate-coding, this work sets the stage for more scalable and efficient SNNs training within resource-constrained environments. Our code is available at https://github.com/Tab-ct/rate-based-backpropagation.
Authors: Rong-Xi Tan, Ke Xue, Shen-Huan Lyu, Haopu Shang, Yao Wang, Yaoyuan Wang, Sheng Fu, Chao Qian
Abstract: Offline model-based optimization (MBO) aims to identify a design that maximizes a black-box function using only a fixed, pre-collected dataset of designs and their corresponding scores. A common approach in offline MBO is to train a regression-based surrogate model by minimizing mean squared error (MSE) and then find the best design within this surrogate model by different optimizers (e.g., gradient ascent). However, a critical challenge is the risk of out-of-distribution errors, i.e., the surrogate model may typically overestimate the scores and mislead the optimizers into suboptimal regions. Prior works have attempted to address this issue in various ways, such as using regularization techniques and ensemble learning to enhance the robustness of the model, but it still remains. In this paper, we argue that regression models trained with MSE are not well-aligned with the primary goal of offline MBO, which is to select promising designs rather than to predict their scores precisely. Notably, if a surrogate model can maintain the order of candidate designs based on their relative score relationships, it can produce the best designs even without precise predictions. To validate it, we conduct experiments to compare the relationship between the quality of the final designs and MSE, finding that the correlation is really very weak. In contrast, a metric that measures order-maintaining quality shows a significantly stronger correlation. Based on this observation, we propose learning a ranking-based model that leverages learning to rank techniques to prioritize promising designs based on their relative scores. We show that the generalization error on ranking loss can be well bounded. Empirical results across diverse tasks demonstrate the superior performance of our proposed ranking-based models than twenty existing methods.
Authors: Youngmok Ha, Yongjoo Kim, Hyun Jae Jang, Seungyeon Lee, Eunji Pak
Abstract: The analysis of biophysical neural networks (BNNs) has been a longstanding focus in computational neuroscience. A central yet unresolved challenge in BNN analysis lies in deciphering the correlations between neuronal and synaptic dynamics, their connectivity patterns, and learning process. To address this, we introduce a novel BNN analysis framework grounded in network representation learning (NRL), which leverages attention scores to uncover intricate correlations between network components and their features. Our framework integrates a new computational graph (CG)-based BNN representation, a bio-inspired graph attention network (BGAN) that enables multiscale correlation analysis across BNN representations, and an extensive BNN dataset. The CG-based representation captures key computational features, information flow, and structural relationships underlying neuronal and synaptic dynamics, while BGAN reflects the compositional structure of neurons, including dendrites, somas, and axons, as well as bidirectional information flows between BNN components. The dataset comprises publicly available models from ModelDB, reconstructed using the Python and standardized in NeuroML format, and is augmented with data derived from canonical neuron and synapse models. To our knowledge, this study is the first to apply an NRL-based approach to the full spectrum of BNNs and their analysis.
Authors: M. Germ\'an-Morales, A. J. Rivera-Rivas, M. J. del Jesus D\'iaz, C. J. Carmona
Abstract: High computational power and the availability of large datasets have supported the development of Foundational Models. They are a new emerging technique widely used in Generative Artificial Intelligence, characterized by their scalability and their use in Transfer Learning. The enormous and heterogeneous amounts of data used in their initial training phase, known as pre-training, give them a higher generalization capacity than any other specific model, constituting a solid base that can be adapted or adjusted to a wide range of tasks, increasing their applicability. This study proposes LLIAM, the Llama Lora-Integrated Autorregresive Model. Low-Rank Adaptations are used to enhance the knowledge of the model with diverse time series datasets, known as the fine-tuning phase. To illustrate the capabilities of our proposal, two sets of experiments have been carried out that obtained favorable and promising results with lower training times than other Deep Learning approaches. With this work, we also encourage the use of available resources (such as these pre-trained models) to avoid unnecessary and costly training, narrowing the gap between the goals of traditional Artificial Intelligence and those specified by the definition of Green Artificial Intelligence.
Authors: Yaxin Du, Rui Ye, Fengting Yuchi, Wanru Zhao, Jingjing Qu, Yanfeng Wang, Siheng Chen
Abstract: By leveraging massively distributed data, federated learning (FL) enables collaborative instruction tuning of large language models (LLMs) in a privacy-preserving way. While FL effectively expands the data quantity, the issue of data quality remains under-explored in the current literature on FL for LLMs. To address this gap, we propose a new framework of federated instruction tuning of LLMs with data quality control (FedDQC), which measures data quality to facilitate the subsequent filtering and hierarchical training processes. Our approach introduces an efficient metric to assess each client's instruction-response alignment (IRA), identifying potentially noisy data through single-shot inference. Low-IRA samples are potentially noisy and filtered to mitigate their negative impacts. To further utilize this IRA value, we propose a quality-aware hierarchical training paradigm, where LLM is progressively fine-tuned from high-IRA to low-IRA data, mirroring the easy-to-hard learning process. We conduct extensive experiments on 4 synthetic and a real-world dataset, and compare our method with baselines adapted from centralized setting. Results show that our method consistently and significantly improves the performance of LLMs trained on mix-quality data in FL.
Authors: Hossein Abdi, Mingfei Sun, Andi Zhang, Samuel Kaski, Wei Pan
Abstract: Training large models with millions or even billions of parameters from scratch incurs substantial computational costs. Parameter Efficient Fine-Tuning (PEFT) methods, particularly Low-Rank Adaptation (LoRA), address this challenge by adapting only a reduced number of parameters to specific tasks with gradient-based optimizers. In this paper, we cast PEFT as an optimal filtering/state estimation problem and present Low-Rank Kalman Optimizer (LoKO) to estimate the optimal trainable parameters in an online manner. We leverage the low-rank decomposition in LoRA to significantly reduce matrix sizes in Kalman iterations and further capitalize on a diagonal approximation of the covariance matrix to effectively decrease computational complexity from quadratic to linear in the number of trainable parameters. Moreover, we discovered that the initialization of the covariance matrix within the Kalman algorithm and the accurate estimation of the observation noise covariance are the keys in this formulation, and we propose robust approaches that work well across a vast range of well-established computer vision and language models. Our results show that LoKO converges with fewer iterations and yields better performance models compared to commonly used optimizers with LoRA in both image classifications and language tasks. Our study opens up the possibility of leveraging the Kalman filter as an effective optimizer for the online fine-tuning of large models.
Authors: Haolin Wang, Xuefeng Liu, Jianwei Niu, Wenkai Guo, Shaojie Tang
Abstract: Federated learning is a distributed machine learning paradigm designed to protect user data privacy, which has been successfully implemented across various scenarios. In traditional federated learning, the entire parameter set of local models is updated and averaged in each training round. Although this full network update method maximizes knowledge acquisition and sharing for each model layer, it prevents the layers of the global model from cooperating effectively to complete the tasks of each client, a challenge we refer to as layer mismatch. This mismatch problem recurs after every parameter averaging, consequently slowing down model convergence and degrading overall performance. To address the layer mismatch issue, we introduce the FedPart method, which restricts model updates to either a single layer or a few layers during each communication round. Furthermore, to maintain the efficiency of knowledge acquisition and sharing, we develop several strategies to select trainable layers in each round, including sequential updating and multi-round cycle training. Through both theoretical analysis and experiments, our findings demonstrate that the FedPart method significantly surpasses conventional full network update strategies in terms of convergence speed and accuracy, while also reducing communication and computational overheads.
Authors: Qingyang Zhang, Qiuxuan Feng, Joey Tianyi Zhou, Yatao Bian, Qinghua Hu, Changqing Zhang
Abstract: Out-of-distribution (OOD) detection is essential for model trustworthiness which aims to sensitively identify semantic OOD samples and robustly generalize for covariate-shifted OOD samples. However, we discover that the superior OOD detection performance of state-of-the-art methods is achieved by secretly sacrificing the OOD generalization ability. Specifically, the classification accuracy of these models could deteriorate dramatically when they encounter even minor noise. This phenomenon contradicts the goal of model trustworthiness and severely restricts their applicability in real-world scenarios. What is the hidden reason behind such a limitation? In this work, we theoretically demystify the ``\textit{sensitive-robust}'' dilemma that lies in many existing OOD detection methods. Consequently, a theory-inspired algorithm is induced to overcome such a dilemma. By decoupling the uncertainty learning objective from a Bayesian perspective, the conflict between OOD detection and OOD generalization is naturally harmonized and a dual-optimal performance could be expected. Empirical studies show that our method achieves superior performance on standard benchmarks. To our best knowledge, this work is the first principled OOD detection method that achieves state-of-the-art OOD detection performance without compromising OOD generalization ability. Our code is available at \href{https://github.com/QingyangZhang/DUL}{https://github.com/QingyangZhang/DUL}.
URLs: https://github.com/QingyangZhang/DUL, https://github.com/QingyangZhang/DUL
Authors: Lech T. Polkowski
Abstract: Rough sets (RS)proved a thriving realm with successes inn many fields of ML and AI. In this note, we expand RS to RM - rough mereology which provides a measurable degree of uncertainty to those areas.
Authors: Chuyang Liu, Tirthankar Roy, Daniel M. Tartakovsky, Dipankar Dwivedi
Abstract: Hydrological models often involve constitutive laws that may not be optimal in every application. We propose to replace such laws with the Kolmogorov-Arnold networks (KANs), a class of neural networks designed to identify symbolic expressions. We demonstrate KAN's potential on the problem of baseflow identification, a notoriously challenging task plagued by significant uncertainty. KAN-derived functional dependencies of the baseflow components on the aridity index outperform their original counterparts. On a test set, they increase the Nash-Sutcliffe Efficiency (NSE) by 67%, decrease the root mean squared error by 30%, and increase the Kling-Gupta efficiency by 24%. This superior performance is achieved while reducing the number of fitting parameters from three to two. Next, we use data from 378 catchments across the continental United States to refine the water-balance equation at the mean-annual scale. The KAN-derived equations based on the refined water balance outperform both the current aridity index model, with up to a 105% increase in NSE, and the KAN-derived equations based on the original water balance. While the performance of our model and tree-based machine learning methods is similar, KANs offer the advantage of simplicity and transparency and require no specific software or computational tools. This case study focuses on the aridity index formulation, but the approach is flexible and transferable to other hydrological processes.
Authors: Nico Wagner, Michael Desmond, Rahul Nair, Zahra Ashktorab, Elizabeth M. Daly, Qian Pan, Mart\'in Santill\'an Cooper, James M. Johnson, Werner Geyer
Abstract: LLM-as-a-Judge is a widely used method for evaluating the performance of Large Language Models (LLMs) across various tasks. We address the challenge of quantifying the uncertainty of LLM-as-a-Judge evaluations. While uncertainty quantification has been well-studied in other domains, applying it effectively to LLMs poses unique challenges due to their complex decision-making capabilities and computational demands. In this paper, we introduce a novel method for quantifying uncertainty designed to enhance the trustworthiness of LLM-as-a-Judge evaluations. The method quantifies uncertainty by analyzing the relationships between generated assessments and possible ratings. By cross-evaluating these relationships and constructing a confusion matrix based on token probabilities, the method derives labels of high or low uncertainty. We evaluate our method across multiple benchmarks, demonstrating a strong correlation between the accuracy of LLM evaluations and the derived uncertainty scores. Our findings suggest that this method can significantly improve the reliability and consistency of LLM-as-a-Judge evaluations.
Authors: Oscar Torres Sanchez, Guilherme Borges, Duarte Raposo, Andr\'e Rodrigues, Fernando Boavida, Jorge S\'a Silva
Abstract: The development of intelligent Industrial Internet of Things (IIoT) systems promises to revolutionize operational and maintenance practices, driving improvements in operational efficiency. Anomaly detection within IIoT architectures plays a crucial role in preventive maintenance and spotting irregularities in industrial components. However, due to limited message and processing capacity, traditional Machine Learning (ML) faces challenges in deploying anomaly detection models in resource-constrained environments like LoRaWAN. On the other hand, Federated Learning (FL) solves this problem by enabling distributed model training, addressing privacy concerns, and minimizing data transmission. This study explores using FL for anomaly detection in industrial and civil construction machinery architectures that use IIoT prototypes with LoRaWAN communication. The process leverages an optimized autoencoder neural network structure and compares federated models with centralized ones. Despite uneven data distribution among machine clients, FL demonstrates effectiveness, with a mean F1 score (of 94.77), accuracy (of 92.30), TNR (of 90.65), and TPR (92.93), comparable to centralized models, considering airtime of trainning messages of 52.8 min. Local model evaluations on each machine highlight adaptability. At the same time, the performed analysis identifies message requirements, minimum training hours, and optimal round/epoch configurations for FL in LoRaWAN, guiding future implementations in constrained industrial environments.
Authors: Aoming Liang, Zhaoyang Mu, Pengxiao Lin, Cong Wang, Mingming Ge, Ling Shao, Dixia Fan, Hao Tang
Abstract: Learning the evolutionary dynamics of Partial Differential Equations (PDEs) is critical in understanding dynamic systems, yet current methods insufficiently learn their representations. This is largely due to the multi-scale nature of the solution, where certain regions exhibit rapid oscillations while others evolve more slowly. This paper introduces a framework of multi-scale and multi-expert (M$^2$M) neural operators designed to simulate and learn PDEs efficiently. We employ a divide-and-conquer strategy to train a multi-expert gated network for the dynamic router policy. Our method incorporates a controllable prior gating mechanism that determines the selection rights of experts, enhancing the model's efficiency. To optimize the learning process, we have implemented a PI (Proportional, Integral) control strategy to adjust the allocation rules precisely. This universal controllable approach allows the model to achieve greater accuracy. We test our approach on benchmark 2D Navier-Stokes equations and provide a custom multi-scale dataset. M$^2$M can achieve higher simulation accuracy and offer improved interpretability compared to baseline methods.
Authors: Jiamian Li
Abstract: Reinforcement learning has achieved remarkable success in perfect information games such as Go and Atari, enabling agents to compete at the highest levels against human players. However, research in reinforcement learning for imperfect information games has been relatively limited due to the more complex game structures and randomness. Traditional methods face challenges in training and improving performance in imperfect information games due to issues like inaccurate Q value estimation and reward sparsity. In this paper, we focus on Uno, an imperfect information game, and aim to address these problems by reducing Q value overestimation and reshaping reward function. We propose a novel algorithm that utilizes Monte Carlo Tree Search to improve the value estimation in Q function. Even though we choose Double Deep Q Learning as the foundational framework in this paper, our method can be generalized and used in any algorithm which needs Q value estimation, such as the Actor-Critic. Additionally, we employ Monte Carlo Tree Search to reshape the reward structure in the game environment. We compared our algorithm with several traditional methods applied to games such as Double Deep Q Learning, Deep Monte Carlo and Neural Fictitious Self Play, and the experiments demonstrate that our algorithm consistently outperforms these approaches, especially as the number of players in Uno increases, indicating a higher level of difficulty.
Authors: Sam McCallum, James Foster
Abstract: Neural ODEs are a recently developed model class that combine the strong model priors of differential equations with the high-capacity function approximation of neural networks. One advantage of Neural ODEs is the potential for memory-efficient training via the continuous adjoint method. However, memory-efficient training comes at the cost of approximate gradients. Therefore, in practice, gradients are often obtained by simply backpropagating through the internal operations of the forward ODE solve - incurring high memory cost. Interestingly, it is possible to construct algebraically reversible ODE solvers that allow for both exact gradients and the memory-efficiency of the continuous adjoint method. Unfortunately, current reversible solvers are low-order and suffer from poor numerical stability. The use of these methods in practice is therefore limited. In this work, we present a class of algebraically reversible solvers that are both high-order and numerically stable. Moreover, any explicit numerical scheme can be made reversible by our method. This construction naturally extends to numerical schemes for Neural CDEs and SDEs.
Authors: Md Kowsher, Md. Shohanur Islam Sobuj, Nusrat Jahan Prottasha, E. Alejandro Alanis, Ozlem Ozmen Garibay, Niloofar Yousefi
Abstract: Time series forecasting remains a challenging task, particularly in the context of complex multiscale temporal patterns. This study presents LLM-Mixer, a framework that improves forecasting accuracy through the combination of multiscale time-series decomposition with pre-trained LLMs (Large Language Models). LLM-Mixer captures both short-term fluctuations and long-term trends by decomposing the data into multiple temporal resolutions and processing them with a frozen LLM, guided by a textual prompt specifically designed for time-series data. Extensive experiments conducted on multivariate and univariate datasets demonstrate that LLM-Mixer achieves competitive performance, outperforming recent state-of-the-art models across various forecasting horizons. This work highlights the potential of combining multiscale analysis and LLMs for effective and scalable time-series forecasting.
Authors: Neeraj Mohan Sushma, Yudou Tian, Harshvardhan Mestha, Nicolo Colombo, David Kappel, Anand Subramoney
Abstract: Deep state-space models (Deep SSMs) have shown capabilities for in-context learning on autoregressive tasks, similar to transformers. However, the architectural requirements and mechanisms enabling this in recurrent networks remain unclear. This study demonstrates that state-space model architectures can perform gradient-based learning and use it for in-context learning. We prove that a single structured state-space model layer, augmented with local self-attention, can reproduce the outputs of an implicit linear model with least squares loss after one step of gradient descent. Our key insight is that the diagonal linear recurrent layer can act as a gradient accumulator, which can be `applied' to the parameters of the implicit regression model. We validate our construction by training randomly initialized augmented SSMs on simple linear regression tasks. The empirically optimized parameters match the theoretical ones, obtained analytically from the implicit model construction. Extensions to multi-step linear and non-linear regression yield consistent results. The constructed SSM encompasses features of modern deep state-space models, with the potential for scalable training and effectiveness even in general tasks. The theoretical construction elucidates the role of local self-attention and multiplicative interactions in recurrent architectures as the key ingredients for enabling the expressive power typical of foundation models.
Authors: Hikaru Shindo, Quentin Delfosse, Devendra Singh Dhami, Kristian Kersting
Abstract: Humans can leverage both symbolic reasoning and intuitive reactions. In contrast, reinforcement learning policies are typically encoded in either opaque systems like neural networks or symbolic systems that rely on predefined symbols and rules. This disjointed approach severely limits the agents' capabilities, as they often lack either the flexible low-level reaction characteristic of neural agents or the interpretable reasoning of symbolic agents. To overcome this challenge, we introduce BlendRL, a neuro-symbolic RL framework that harmoniously integrates both paradigms within RL agents that use mixtures of both logic and neural policies. We empirically demonstrate that BlendRL agents outperform both neural and symbolic baselines in standard Atari environments, and showcase their robustness to environmental changes. Additionally, we analyze the interaction between neural and symbolic policies, illustrating how their hybrid use helps agents overcome each other's limitations.
Authors: Nina Wiedemann, Th\'eo Uscidda, Martin Raubal
Abstract: Prediction problems in geographic information science and transportation are frequently motivated by the possibility to enhance operational efficiency. Examples range from predicting car sharing demand for optimizing relocation to forecasting traffic congestion for navigation purposes. However, conventional accuracy metrics do not account for the spatial distribution of predictions errors, despite its relevance for operations. We put forward Optimal Transport (OT) as a spatial evaluation metric and loss function. The proposed OT metric assesses the utility of spatial prediction models in terms of the relocation costs caused by prediction errors. In experiments on real and synthetic data, we demonstrate that 1) the spatial distribution of the prediction errors is relevant in many applications and can be translated to real-world costs, 2) in contrast to other metrics, OT reflects these spatial costs, and 3) OT metrics improve comparability across spatial and temporal scales. Finally, we advocate for leveraging OT as a loss function in neural networks to improve the spatial correctness of predictions. This approach not only aligns evaluation in GeoAI with operational considerations, but also signifies a step forward in refining predictions within geospatial applications. To facilitate the adoption of OT in GIS, we provide code and tutorials at https://github.com/mie-lab/geospatialOT.
Authors: Yunfan Xiong, Ruoyu Zhang, Yanzeng Li, Tianhao Wu, Lei Zou
Abstract: While speculative decoding has recently appeared as a promising direction for accelerating the inference of large language models (LLMs), the speedup and scalability are strongly bounded by the token acceptance rate. Prevalent methods usually organize predicted tokens as independent chains or fixed token trees, which fails to generalize to diverse query distributions. In this paper, we propose DySpec, a faster speculative decoding algorithm with a novel dynamic token tree structure. We begin by bridging the draft distribution and acceptance rate from intuitive and empirical clues, and successfully show that the two variables are strongly correlated. Based on this, we employ a greedy strategy to dynamically expand the token tree at run time. Theoretically, we show that our method can achieve optimal results under mild assumptions. Empirically, DySpec yields a higher acceptance rate and speedup than fixed trees. DySpec can drastically improve the throughput and reduce the latency of token generation across various data distribution and model sizes, which significantly outperforms strong competitors, including Specinfer and Sequoia. Under low temperature setting, DySpec can improve the throughput up to 9.1$\times$ and reduce the latency up to 9.4$\times$ on Llama2-70B. Under high temperature setting, DySpec can also improve the throughput up to 6.21$\times$, despite the increasing difficulty of speculating more than one token per step for draft model.
Authors: Sujai Hiremath, Kyra Gan, Promit Ghosal
Abstract: We address the challenge of causal discovery in structural equation models with additive noise without imposing additional assumptions on the underlying data-generating process. We introduce local search in additive noise model (LoSAM), which generalizes an existing nonlinear method that leverages local causal substructures to the general additive noise setting, allowing for both linear and nonlinear causal mechanisms. We show that LoSAM achieves polynomial runtime, and improves runtime and efficiency by exploiting new substructures to minimize the conditioning set at each step. Further, we introduce a variant of LoSAM, LoSAM-UC, that is robust to unmeasured confounding among roots, a property that is often not satisfied by functional-causal-model-based methods. We numerically demonstrate the utility of LoSAM, showing that it outperforms existing benchmarks.
Authors: Bishal Thapaliya, Anh Nguyen, Yao Lu, Tian Xie, Igor Grudetskyi, Fudong Lin, Antonios Valkanas, Jingyu Liu, Deepayan Chakraborty, Bilel Fehri
Abstract: Classifying nodes in a graph is a common problem. The ideal classifier must adapt to any imbalances in the class distribution. It must also use information in the clustering structure of real-world graphs. Existing Graph Neural Networks (GNNs) have not addressed both problems together. We propose the Enhanced Cluster-aware Graph Network (ECGN), a novel method that addresses these issues by integrating cluster-specific training with synthetic node generation. Unlike traditional GNNs that apply the same node update process for all nodes, ECGN learns different aggregations for different clusters. We also use the clusters to generate new minority-class nodes in a way that helps clarify the inter-class decision boundary. By combining cluster-aware embeddings with a global integration step, ECGN enhances the quality of the resulting node embeddings. Our method works with any underlying GNN and any cluster generation technique. Experimental results show that ECGN consistently outperforms its closest competitors by up to 11% on some widely studied benchmark datasets.
Authors: Abhinav Menon, Manish Shrivastava, David Krueger, Ekdeep Singh Lubana
Abstract: Autoencoders have been used for finding interpretable and disentangled features underlying neural network representations in both image and text domains. While the efficacy and pitfalls of such methods are well-studied in vision, there is a lack of corresponding results, both qualitative and quantitative, for the text domain. We aim to address this gap by training sparse autoencoders (SAEs) on a synthetic testbed of formal languages. Specifically, we train SAEs on the hidden representations of models trained on formal languages (Dyck-2, Expr, and English PCFG) under a wide variety of hyperparameter settings, finding interpretable latents often emerge in the features learned by our SAEs. However, similar to vision, we find performance turns out to be highly sensitive to inductive biases of the training pipeline. Moreover, we show latents correlating to certain features of the input do not always induce a causal impact on model's computation. We thus argue that causality has to become a central target in SAE training: learning of causal features should be incentivized from the ground-up. Motivated by this, we propose and perform preliminary investigations for an approach that promotes learning of causally relevant features in our formal language setting.
Authors: Stephane Bersier, Xinyi Chen-Lin
Abstract: Despite the wide variety of input types in machine learning, this diversity is often not fully reflected in their representations or model architectures, leading to inefficiencies throughout a model's lifecycle. This paper introduces an algebraic approach to constructing input-encoding architectures that properly account for the data's structure, providing a step toward achieving more typeful machine learning.
Authors: Wei Shen, Ruida Zhou, Jing Yang, Cong Shen
Abstract: While transformers have demonstrated impressive capacities for in-context learning (ICL) in practice, theoretical understanding of the underlying mechanism enabling transformers to perform ICL is still in its infant stage. This work aims to theoretically study the training dynamics of transformers for in-context classification tasks. We demonstrate that, for in-context classification of Gaussian mixtures under certain assumptions, a single-layer transformer trained via gradient descent converges to a globally optimal model at a linear rate. We further quantify the impact of the training and testing prompt lengths on the ICL inference error of the trained transformer. We show that when the lengths of training and testing prompts are sufficiently large, the prediction of the trained transformer approaches the Bayes-optimal classifier. Experimental results corroborate the theoretical findings.
Authors: Amit Arnold Levy, Mor Geva
Abstract: Large language models (LLMs) frequently make errors when handling even simple numerical problems, such as comparing two small numbers. A natural hypothesis is that these errors stem from how LLMs represent numbers, and specifically, whether their representations of numbers capture their numeric values. We tackle this question from the observation that LLM errors on numerical tasks are often distributed across \textit{the digits} of the answer rather than normally around \textit{its numeric value}. Through a series of probing experiments and causal interventions, we show that LLMs internally represent numbers with individual circular representations per-digit in base 10. This digit-wise representation, as opposed to a value representation, sheds light on the error patterns of models on tasks involving numerical reasoning and could serve as a basis for future studies on analyzing numerical mechanisms in LLMs.
Authors: Zhe Li, Xiangfei Qiu, Peng Chen, Yihang Wang, Hanyin Cheng, Yang Shu, Jilin Hu, Chenjuan Guo, Aoying Zhou, Qingsong Wen, Christian S. Jensen, Bin Yang
Abstract: Time Series Forecasting (TSF) is key functionality in numerous fields, including in finance, weather services, and energy management. While TSF methods are emerging these days, many of them require domain-specific data collection and model training and struggle with poor generalization performance on new domains. Foundation models aim to overcome this limitation. Pre-trained on large-scale language or time series data, they exhibit promising inferencing capabilities in new or unseen data. This has spurred a surge in new TSF foundation models. We propose a new benchmark, FoundTS, to enable thorough and fair evaluation and comparison of such models. FoundTS covers a variety of TSF foundation models, including those based on large language models and those pretrained on time series. Next, FoundTS supports different forecasting strategies, including zero-shot, few-shot, and full-shot, thereby facilitating more thorough evaluations. Finally, FoundTS offers a pipeline that standardizes evaluation processes such as dataset splitting, loading, normalization, and few-shot sampling, thereby facilitating fair evaluations. Building on this, we report on an extensive evaluation of TSF foundation models on a broad range of datasets from diverse domains and with different statistical characteristics. Specifically, we identify pros and cons and inherent limitations of existing foundation models, and we identify directions for future model design. We make our code and datasets available at https://anonymous.4open.science/r/FoundTS-C2B0.
Authors: Yiding Jiang, Allan Zhou, Zhili Feng, Sadhika Malladi, J. Zico Kolter
Abstract: The composition of pretraining data is a key determinant of foundation models' performance, but there is no standard guideline for allocating a limited computational budget across different data sources. Most current approaches either rely on extensive experiments with smaller models or dynamic data adjustments that also require proxy models, both of which significantly increase the workflow complexity and computational overhead. In this paper, we introduce Adaptive Data Optimization (ADO), an algorithm that optimizes data distributions in an online fashion, concurrent with model training. Unlike existing techniques, ADO does not require external knowledge, proxy models, or modifications to the model update. Instead, ADO uses per-domain scaling laws to estimate the learning potential of each domain during training and adjusts the data mixture accordingly, making it more scalable and easier to integrate. Experiments demonstrate that ADO can achieve comparable or better performance than prior methods while maintaining computational efficiency across different computation scales, offering a practical solution for dynamically adjusting data distribution without sacrificing flexibility or increasing costs. Beyond its practical benefits, ADO also provides a new perspective on data collection strategies via scaling laws.
Authors: Ayush Jain, Norio Kosaka, Xinhu Li, Kyung-Min Kim, Erdem B{\i}y{\i}k, Joseph J. Lim
Abstract: In reinforcement learning, off-policy actor-critic approaches like DDPG and TD3 are based on the deterministic policy gradient. Herein, the Q-function is trained from off-policy environment data and the actor (policy) is trained to maximize the Q-function via gradient ascent. We observe that in complex tasks like dexterous manipulation and restricted locomotion, the Q-value is a complex function of action, having several local optima or discontinuities. This poses a challenge for gradient ascent to traverse and makes the actor prone to get stuck at local optima. To address this, we introduce a new actor architecture that combines two simple insights: (i) use multiple actors and evaluate the Q-value maximizing action, and (ii) learn surrogates to the Q-function that are simpler to optimize with gradient-based methods. We evaluate tasks such as restricted locomotion, dexterous manipulation, and large discrete-action space recommender systems and show that our actor finds optimal actions more frequently and outperforms alternate actor architectures.
Authors: Leshem Choshen, Yang Zhang, Jacob Andreas
Abstract: Scaling laws predict the loss of a target machine learning model by extrapolating from easier-to-train models with fewer parameters or smaller training sets. This provides an efficient way for practitioners and researchers alike to compare pretraining decisions involving optimizers, datasets, and model architectures. Despite the widespread use of scaling laws to model the dynamics of language model training, there has been little work on understanding how to best estimate and interpret them. We collect (and release) a large-scale dataset containing losses and downstream evaluations for 485 previously published pretrained models. We use these to estimate more than 1000 scaling laws, then derive a set of best practices for estimating scaling laws in new model families. We find that fitting scaling laws to intermediate checkpoints of training runs (and not just their final losses) substantially improves accuracy, and that -- all else equal -- estimates of performance are generally most accurate when derived from other models of similar sizes. However, because there is a significant degree of variability across model seeds, training multiple small models is sometimes more useful than training a single large one. Moreover, while different model families differ scaling behavior, they are often similar enough that a target model's behavior can be predicted from a single model with the same architecture, along with scaling parameter estimates derived from other model families.
Authors: Siyeop Yoon, Yujin Oh, Xiang Li, Yi Xin, Maurizio Cereda, Quanzheng Li
Abstract: Acute respiratory distress syndrome (ARDS) is a severe condition characterized by lung inflammation and respiratory failure, with a high mortality rate of approximately 40%. Traditional imaging methods, such as chest X-rays, provide only two-dimensional views, limiting their effectiveness in fully assessing lung pathology. Three-dimensional (3D) computed tomography (CT) offers a more comprehensive visualization, enabling detailed analysis of lung aeration, atelectasis, and the effects of therapeutic interventions. However, the routine use of CT in ARDS management is constrained by practical challenges and risks associated with transporting critically ill patients to remote scanners. In this study, we synthesize high-fidelity 3D lung CT from 2D generated X-ray images with associated physiological parameters using a score-based 3D residual diffusion model. Our preliminary results demonstrate that this approach can produce high-quality 3D CT images that are validated with ground truth, offering a promising solution for enhancing ARDS management.
Authors: Zhangqi Duan, Nigel Fernandez, Alexander Hicks, Andrew Lan
Abstract: Open-ended coding tasks, which ask students to construct programs according to certain specifications, are common in computer science education. Student modeling can be challenging since their open-ended nature means that student code can be diverse. Traditional knowledge tracing (KT) models that only analyze response correctness may not fully capture nuances in student knowledge from student code. In this paper, we introduce Test case-Informed Knowledge Tracing for Open-ended Coding (TIKTOC), a framework to simultaneously analyze and predict both open-ended student code and whether the code passes each test case. We augment the existing CodeWorkout dataset with the test cases used for a subset of the open-ended coding questions, and propose a multi-task learning KT method to simultaneously analyze and predict 1) whether a student's code submission passes each test case and 2) the student's open-ended code, using a large language model as the backbone. We quantitatively show that these methods outperform existing KT methods for coding that only use the overall score a code submission receives. We also qualitatively demonstrate how test case information, combined with open-ended code, helps us gain fine-grained insights into student knowledge.
Authors: Zhidong Gao, Zhenxiao Zhang, Yu Zhang, Tongnian Wang, Yanmin Gong, Yuanxiong Guo
Abstract: Federated learning (FL) enables edge devices to collaboratively train a machine learning model without sharing their raw data. Due to its privacy-protecting benefits, FL has been deployed in many real-world applications. However, deploying FL over mobile edge networks with constrained resources such as power, bandwidth, and computation suffers from high training latency and low model accuracy, particularly under data and system heterogeneity. In this paper, we investigate the optimal client scheduling and resource allocation for FL over mobile edge networks under resource constraints and uncertainty to minimize the training latency while maintaining the model accuracy. Specifically, we first analyze the impact of client sampling on model convergence in FL and formulate a stochastic optimization problem that captures the trade-off between the running time and model performance under heterogeneous and uncertain system resources. To solve the formulated problem, we further develop an online control scheme based on Lyapunov-based optimization for client sampling and resource allocation without requiring the knowledge of future dynamics in the FL system. Extensive experimental results demonstrate that the proposed scheme can improve both the training latency and resource efficiency compared with the existing schemes.
Authors: Di Zhang, Bowen Lv, Hai Zhang, Feifan Yang, Junqiao Zhao, Hang Yu, Chang Huang, Hongtu Zhou, Chen Ye, Changjun Jiang
Abstract: A primary challenge for visual-based Reinforcement Learning (RL) is to generalize effectively across unseen environments. Although previous studies have explored different auxiliary tasks to enhance generalization, few adopt image reconstruction due to concerns about exacerbating overfitting to task-irrelevant features during training. Perceiving the pre-eminence of image reconstruction in representation learning, we propose SMG (Separated Models for Generalization), a novel approach that exploits image reconstruction for generalization. SMG introduces two model branches to extract task-relevant and task-irrelevant representations separately from visual observations via cooperatively reconstruction. Built upon this architecture, we further emphasize the importance of task-relevant features for generalization. Specifically, SMG incorporates two additional consistency losses to guide the agent's focus toward task-relevant areas across different scenarios, thereby achieving free from overfitting. Extensive experiments in DMC demonstrate the SOTA performance of SMG in generalization, particularly excelling in video-background settings. Evaluations on robotic manipulation tasks further confirm the robustness of SMG in real-world applications.
Authors: Heyuan Huang, Xingyu Lou, Chaochao Chen, Pengxiang Cheng, Yue Xin, Chengwei He, Xiang Liu, Jun Wang
Abstract: Cross-Domain Recommendation (CDR) have received widespread attention due to their ability to utilize rich information across domains. However, most existing CDR methods assume an ideal static condition that is not practical in industrial recommendation systems (RS). Therefore, simply applying existing CDR methods in the industrial RS environment may lead to low effectiveness and efficiency. To fill this gap, we propose DIIT, an end-to-end Domain-Invariant Information Transfer method for industrial cross-domain recommendation. Specifically, We first simulate the industrial RS environment that maintains respective models in multiple domains, each of them is trained in the incremental mode. Then, for improving the effectiveness, we design two extractors to fully extract domain-invariant information from the latest source domain models at the domain level and the representation level respectively. Finally, for improving the efficiency, we design a migrator to transfer the extracted information to the latest target domain model, which only need the target domain model for inference. Experiments conducted on one production dataset and two public datasets verify the effectiveness and efficiency of DIIT.
Authors: Yifan Gong, Yushu Wu, Zheng Zhan, Pu Zhao, Liangkai Liu, Chao Wu, Xulong Tang, Yanzhi Wang
Abstract: Two-stage object detectors exhibit high accuracy and precise localization, especially for identifying small objects that are favorable for various edge applications. However, the high computation costs associated with two-stage detection methods cause more severe thermal issues on edge devices, incurring dynamic runtime frequency change and thus large inference latency variations. Furthermore, the dynamic number of proposals in different frames leads to various computations over time, resulting in further latency variations. The significant latency variations of detectors on edge devices can harm user experience and waste hardware resources. To avoid thermal throttling and provide stable inference speed, we propose Lotus, a novel framework that is tailored for two-stage detectors to dynamically scale CPU and GPU frequencies jointly in an online manner based on deep reinforcement learning (DRL). To demonstrate the effectiveness of Lotus, we implement it on NVIDIA Jetson Orin Nano and Mi 11 Lite mobile platforms. The results indicate that Lotus can consistently and significantly reduce latency variation, achieve faster inference, and maintain lower CPU and GPU temperatures under various settings.
Authors: Haozhou Pang, Tianwei Ding, Lanshan He, Qi Gan
Abstract: In this work, we present LLM Gesticulator, an LLM-based audio-driven co-speech gesture generation framework that synthesizes full-body animations that are rhythmically aligned with the input audio while exhibiting natural movements and editability. Compared to previous work, our model demonstrates substantial scalability. As the size of the backbone LLM model increases, our framework shows proportional improvements in evaluation metrics (a.k.a. scaling law). Our method also exhibits strong controllability where the content, style of the generated gestures can be controlled by text prompt. To the best of our knowledge, LLM gesticulator is the first work that use LLM on the co-speech generation task. Evaluation with existing objective metrics and user studies indicate that our framework outperforms prior works.
Authors: Yew Ken Chia, Guizhen Chen, Weiwen Xu, Luu Anh Tuan, Soujanya Poria, Lidong Bing
Abstract: Advanced models such as OpenAI o1 exhibit impressive problem-solving capabilities through step-by-step reasoning. However, they may still falter on more complex problems, making errors that disrupt their reasoning paths. We attribute this to the expansive solution space, where each step has the risk of diverging into mistakes. To enhance language model reasoning, we introduce a specialized training framework called Reasoning Paths Optimization (RPO), which enables learning to reason and explore from diverse paths. Our approach encourages favorable branches at each reasoning step while penalizing unfavorable ones, enhancing the model's overall problem-solving performance. Reasoning Paths Optimization does not rely on large-scale human-annotated rationales or outputs from closed-source models, making it scalable and data-efficient. We focus on multi-step reasoning tasks, such as math word problems and science-based exam questions. The experiments demonstrate that our framework significantly enhances the reasoning performance of large language models, with up to 3.1% and 4.3% improvement on GSM8K and MMLU (STEM) respectively. Our data and code can be found at https://reasoning-paths.github.io.
Authors: Jianwei Li, Jung-Eun Kim
Abstract: As large language models (LLMs) are overwhelmingly more and more integrated into various applications, ensuring they generate safe and aligned responses is a pressing need. Previous research on alignment has largely focused on general instruction-following but has often overlooked the unique properties and challenges of safety alignment, such as the brittleness of safety mechanisms. To bridge the gap, we propose the Superficial Safety Alignment Hypothesis (SSAH), which posits that safety alignment should teach an otherwise unsafe model to choose the correct reasoning direction - interpreted as a specialized binary classification task - and incorporate a refusal mechanism with multiple reserved fallback options. Furthermore, through SSAH, we hypothesize that safety guardrails in LLMs can be established by just a small number of essential components. To verify this, we conduct an ablation study and successfully identify four types of attribute-critical components in safety-aligned LLMs: Exclusive Safety Unit (ESU), Exclusive Utility Unit (EUU), Complex Unit (CU), and Redundant Unit (RU). Our findings show that freezing certain safety-critical components 7.5\% during fine-tuning allows the model to retain its safety attributes while adapting to new tasks. Additionally, we show that leveraging redundant units 20\% in the pre-trained model as an ``alignment budget'' can effectively minimize the alignment tax while achieving the alignment goal. All considered, this paper concludes that the atomic functional unit for safety in LLMs is at the neuron level and underscores that safety alignment should not be complicated. We believe this work contributes to the foundation of efficient and scalable safety alignment for future LLMs.
Authors: Yang Ba, Michelle V. Mancenido, Rong Pan
Abstract: As machine learning models continue to swiftly advance, calibrating their performance has become a major concern prior to practical and widespread implementation. Most existing calibration methods often negatively impact model accuracy due to the lack of diversity of validation data, resulting in reduced generalizability. To address this, we propose a calibration method that incorporates synthetic data without compromising accuracy. We derive the expected calibration error (ECE) bound using the Probably Approximately Correct (PAC) learning framework. Large language models (LLMs), known for their ability to mimic real data and generate text with mixed class labels, are utilized as a synthetic data generation strategy to lower the ECE bound and improve model accuracy on real test data. Additionally, we propose data generation mechanisms for efficient calibration. Testing our method on four different natural language processing tasks, we observed an average up to 34\% increase in accuracy and 33\% decrease in ECE.
Authors: Ryota Tozuka, Hisashi Johno, Akitomo Amakawa, Junichi Sato, Mizuki Muto, Shoichiro Seki, Atsushi Komaba, Hiroshi Onishi
Abstract: Purpose: In radiology, large language models (LLMs), including ChatGPT, have recently gained attention, and their utility is being rapidly evaluated. However, concerns have emerged regarding their reliability in clinical applications due to limitations such as hallucinations and insufficient referencing. To address these issues, we focus on the latest technology, retrieval-augmented generation (RAG), which enables LLMs to reference reliable external knowledge (REK). Specifically, this study examines the utility and reliability of a recently released RAG-equipped LLM (RAG-LLM), NotebookLM, for staging lung cancer. Materials and methods: We summarized the current lung cancer staging guideline in Japan and provided this as REK to NotebookLM. We then tasked NotebookLM with staging 100 fictional lung cancer cases based on CT findings and evaluated its accuracy. For comparison, we performed the same task using a gold-standard LLM, GPT-4 Omni (GPT-4o), both with and without the REK. Results: NotebookLM achieved 86% diagnostic accuracy in the lung cancer staging experiment, outperforming GPT-4o, which recorded 39% accuracy with the REK and 25% without it. Moreover, NotebookLM demonstrated 95% accuracy in searching reference locations within the REK. Conclusion: NotebookLM successfully performed lung cancer staging by utilizing the REK, demonstrating superior performance compared to GPT-4o. Additionally, it provided highly accurate reference locations within the REK, allowing radiologists to efficiently evaluate the reliability of NotebookLM's responses and detect possible hallucinations. Overall, this study highlights the potential of NotebookLM, a RAG-LLM, in image diagnosis.
Authors: Rana Muhammad Shahroz Khan, Pingzhi Li, Sukwon Yun, Zhenyu Wang, Shahriar Nirjon, Chau-Wai Wong, Tianlong Chen
Abstract: As large language models (LLMs) increasingly shape the AI landscape, fine-tuning pretrained models has become more popular than in the pre-LLM era for achieving optimal performance in domain-specific tasks. However, pretrained LLMs such as ChatGPT are periodically evolved, i.e., model parameters are frequently updated), making it challenging for downstream users with limited resources to keep up with fine-tuning the newest LLMs for their domain application. Even though fine-tuning costs have nowadays been reduced thanks to the innovations of parameter-efficient fine-tuning such as LoRA, not all downstream users have adequate computing for frequent personalization. Moreover, access to fine-tuning datasets, particularly in sensitive domains such as healthcare, could be time-restrictive, making it crucial to retain the knowledge encoded in earlier fine-tuned rounds for future adaptation. In this paper, we present PortLLM, a training-free framework that (i) creates an initial lightweight model update patch to capture domain-specific knowledge, and (ii) allows a subsequent seamless plugging for the continual personalization of evolved LLM at minimal cost. Our extensive experiments cover seven representative datasets, from easier question-answering tasks {BoolQ, SST2} to harder reasoning tasks {WinoGrande, GSM8K}, and models including {Mistral-7B, Llama2, Llama3.1, and Gemma2}, validating the portability of our designed model patches and showcasing the effectiveness of our proposed framework. For instance, PortLLM achieves comparable performance to LoRA fine-tuning with reductions of up to 12.2x in GPU memory usage. Finally, we provide theoretical justifications to understand the portability of our model update patches, which offers new insights into the theoretical dimension of LLMs' personalization.
Authors: Hamed Sajadinia, Ali Aghdaei, Zhuo Feng
Abstract: State-of-the-art hypergraph partitioners utilize a multilevel paradigm to construct progressively coarser hypergraphs across multiple layers, guiding cut refinements at each level of the hierarchy. Traditionally, these partitioners employ heuristic methods for coarsening and do not consider the structural features of hypergraphs. In this work, we introduce a multilevel spectral framework, SHyPar, for partitioning large-scale hypergraphs by leveraging hyperedge effective resistances and flow-based community detection techniques. Inspired by the latest theoretical spectral clustering frameworks, such as HyperEF and HyperSF, SHyPar aims to decompose large hypergraphs into multiple subgraphs with few inter-partition hyperedges (cut size). A key component of SHyPar is a flow-based local clustering scheme for hypergraph coarsening, which incorporates a max-flow-based algorithm to produce clusters with substantially improved conductance. Additionally, SHyPar utilizes an effective resistance-based rating function for merging nodes that are strongly connected (coupled). Compared with existing state-of-the-art hypergraph partitioning methods, our extensive experimental results on real-world VLSI designs demonstrate that SHyPar can more effectively partition hypergraphs, achieving state-of-the-art solution quality.
Authors: Zhenyu Xu, Kun Zhang, Victor S. Sheng
Abstract: The increasing use of Large Language Models (LLMs) for generating highly coherent and contextually relevant text introduces new risks, including misuse for unethical purposes such as disinformation or academic dishonesty. To address these challenges, we propose FreqMark, a novel watermarking technique that embeds detectable frequency-based watermarks in LLM-generated text during the token sampling process. The method leverages periodic signals to guide token selection, creating a watermark that can be detected with Short-Time Fourier Transform (STFT) analysis. This approach enables accurate identification of LLM-generated content, even in mixed-text scenarios with both human-authored and LLM-generated segments. Our experiments demonstrate the robustness and precision of FreqMark, showing strong detection capabilities against various attack scenarios such as paraphrasing and token substitution. Results show that FreqMark achieves an AUC improvement of up to 0.98, significantly outperforming existing detection methods.
Authors: Guoxiong Gao, Yutong Wang, Jiedong Jiang, Qi Gao, Zihan Qin, Tianyi Xu, Bin Dong
Abstract: Verifiable formal languages like Lean have profoundly impacted mathematical reasoning, particularly through the use of large language models (LLMs) for automated reasoning. A significant challenge in training LLMs for these formal languages is the lack of parallel datasets that align natural language with formal language proofs. To address this challenge, this paper introduces a novel framework for translating the Mathlib4 corpus (a unified library of mathematics in formal language Lean 4) into natural language. Building upon this, we employ a dual augmentation strategy that combines tactic-based and informal-based approaches, leveraging the Lean-jixia system, a Lean 4 analyzer. We present the results of this pipeline on Mathlib4 as Herald (Hierarchy and Retrieval-based Translated Lean Dataset). We also propose the Herald Translator, which is fine-tuned on Herald. Herald translator achieves a 93.2% accuracy (Pass@128) on formalizing statements in the miniF2F-test and a 22.5% accuracy on our internal graduate-level textbook dataset, outperforming InternLM2-Math-Plus-7B (74.0% and 7.5%) and TheoremLlama (50.1% and 4.0%). Furthermore, we propose a section-level translation framework for real-world applications. As a direct application of Herald translator, we have successfully translated a template section in the Stack project, marking a notable progress in the automatic formalization of graduate-level mathematical literature. Our model, along with the datasets, will be open-sourced to the public soon.
Authors: Hengxiang Zhang, Songxin Zhang, Bingyi Jing, Hongxin Wei
Abstract: In the era of large language models (LLMs), detecting pretraining data has been increasingly important due to concerns about fair evaluation and ethical risks. Current methods differentiate members and non-members by designing scoring functions, like Perplexity and Min-k%. However, the diversity and complexity of training data magnifies the difficulty of distinguishing, leading to suboptimal performance in detecting pretraining data. In this paper, we first explore the benefits of unseen data, which can be easily collected after the release of the LLM. We find that the perplexities of LLMs perform differently for members and non-members, after fine-tuning with a small amount of previously unseen data. In light of this, we introduce a novel and effective method termed Fine-tuned Score Deviation (FSD), which improves the performance of current scoring functions for pretraining data detection. In particular, we propose to measure the deviation distance of current scores after fine-tuning on a small amount of unseen data within the same domain. In effect, using a few unseen data can largely decrease the scores of all non-members, leading to a larger deviation distance than members. Extensive experiments demonstrate the effectiveness of our method, significantly improving the AUC score on common benchmark datasets across various models.
Authors: Leonardo Sabattini, Annalisa Coriolano, Corneel Casert, Stiven Forti, Edward S. Barnard, Fabio Beltram, Massimiliano Pontil, Stephen Whitelam, Camilla Coletti, Antonio Rossi
Abstract: Two-dimensional (2D) materials are poised to revolutionize current solid-state technology with their extraordinary properties. Yet, the primary challenge remains their scalable production. While there have been significant advancements, much of the scientific progress has depended on the exfoliation of materials, a method that poses severe challenges for large-scale applications. With the advent of artificial intelligence (AI) in materials science, innovative synthesis methodologies are now on the horizon. This study explores the forefront of autonomous materials synthesis using an artificial neural network (ANN) trained by evolutionary methods, focusing on the efficient production of graphene. Our approach demonstrates that a neural network can iteratively and autonomously learn a time-dependent protocol for the efficient growth of graphene, without requiring pretraining on what constitutes an effective recipe. Evaluation criteria are based on the proximity of the Raman signature to that of monolayer graphene: higher scores are granted to outcomes whose spectrum more closely resembles that of an ideal continuous monolayer structure. This feedback mechanism allows for iterative refinement of the ANN's time-dependent synthesis protocols, progressively improving sample quality. Through the advancement and application of AI methodologies, this work makes a substantial contribution to the field of materials engineering, fostering a new era of innovation and efficiency in the synthesis process.
Authors: Sihan Liu, Christopher Ye
Abstract: Uniformity testing is arguably one of the most fundamental distribution testing problems. Given sample access to an unknown distribution $\mathbf{p}$ on $[n]$, one must decide if $\mathbf{p}$ is uniform or $\varepsilon$-far from uniform (in total variation distance). A long line of work established that uniformity testing has sample complexity $\Theta(\sqrt{n}\varepsilon^{-2})$. However, when the input distribution is neither uniform nor far from uniform, known algorithms may have highly non-replicable behavior. Consequently, if these algorithms are applied in scientific studies, they may lead to contradictory results that erode public trust in science. In this work, we revisit uniformity testing under the framework of algorithmic replicability [STOC '22], requiring the algorithm to be replicable under arbitrary distributions. While replicability typically incurs a $\rho^{-2}$ factor overhead in sample complexity, we obtain a replicable uniformity tester using only $\tilde{O}(\sqrt{n} \varepsilon^{-2} \rho^{-1})$ samples. To our knowledge, this is the first replicable learning algorithm with (nearly) linear dependence on $\rho$. Lastly, we consider a class of ``symmetric" algorithms [FOCS '00] whose outputs are invariant under relabeling of the domain $[n]$, which includes all existing uniformity testers (including ours). For this natural class of algorithms, we prove a nearly matching sample complexity lower bound for replicable uniformity testing.
Authors: Qingyang Zhang, Yatao Bian, Xinke Kong, Peilin Zhao, Changqing Zhang
Abstract: Machine learning models must continuously self-adjust themselves for novel data distribution in the open world. As the predominant principle, entropy minimization (EM) has been proven to be a simple yet effective cornerstone in existing test-time adaption (TTA) methods. While unfortunately its fatal limitation (i.e., overconfidence) tends to result in model collapse. For this issue, we propose to Conservatively Minimize the Entropy (COME), which is a simple drop-in replacement of traditional EM to elegantly address the limitation. In essence, COME explicitly models the uncertainty by characterizing a Dirichlet prior distribution over model predictions during TTA. By doing so, COME naturally regularizes the model to favor conservative confidence on unreliable samples. Theoretically, we provide a preliminary analysis to reveal the ability of COME in enhancing the optimization stability by introducing a data-adaptive lower bound on the entropy. Empirically, our method achieves state-of-the-art performance on commonly used benchmarks, showing significant improvements in terms of classification accuracy and uncertainty estimation under various settings including standard, life-long and open-world TTA, i.e., up to $34.5\%$ improvement on accuracy and $15.1\%$ on false positive rate.
Authors: Farinaz Mostajeran, Salah A Faroughi
Abstract: Multilayer perceptron (MLP) networks are predominantly used to develop data-driven constitutive models for granular materials. They offer a compelling alternative to traditional physics-based constitutive models in predicting nonlinear responses of these materials, e.g., elasto-plasticity, under various loading conditions. To attain the necessary accuracy, MLPs often need to be sufficiently deep or wide, owing to the curse of dimensionality inherent in these problems. To overcome this limitation, we present an elasto-plasticity informed Chebyshev-based Kolmogorov-Arnold network (EPi-cKAN) in this study. This architecture leverages the benefits of KANs and augmented Chebyshev polynomials, as well as integrates physical principles within both the network structure and the loss function. The primary objective of EPi-cKAN is to provide an accurate and generalizable function approximation for non-linear stress-strain relationships, using fewer parameters compared to standard MLPs. To evaluate the efficiency, accuracy, and generalization capabilities of EPi-cKAN in modeling complex elasto-plastic behavior, we initially compare its performance with other cKAN-based models, which include purely data-driven parallel and serial architectures. Furthermore, to differentiate EPi-cKAN's distinct performance, we also compare it against purely data-driven and physics-informed MLP-based methods. Lastly, we test EPi-cKAN's ability to predict blind strain-controlled paths that extend beyond the training data distribution to gauge its generalization and predictive capabilities. Our findings indicate that, even with limited data and fewer parameters compared to other approaches, EPi-cKAN provides superior accuracy in predicting stress components and demonstrates better generalization when used to predict sand elasto-plastic behavior under blind triaxial axisymmetric strain-controlled loading paths.
Authors: Kyungeun Lee, Wonjong Rhee
Abstract: Mutual Information (MI) is a fundamental metric for quantifying dependency between two random variables. When we can access only the samples, but not the underlying distribution functions, we can evaluate MI using sample-based estimators. Assessment of such MI estimators, however, has almost always relied on analytical datasets including Gaussian multivariates. Such datasets allow analytical calculations of the true MI values, but they are limited in that they do not reflect the complexities of real-world datasets. This study introduces a comprehensive benchmark suite for evaluating neural MI estimators on unstructured datasets, specifically focusing on images and texts. By leveraging same-class sampling for positive pairing and introducing a binary symmetric channel trick, we show that we can accurately manipulate true MI values of real-world datasets. Using the benchmark suite, we investigate seven challenging scenarios, shedding light on the reliability of neural MI estimators for unstructured datasets.
Authors: Max Hamilton, Christian Lange, Elijah Cole, Alexander Shepard, Samuel Heinrich, Oisin Mac Aodha, Grant Van Horn, Subhransu Maji
Abstract: Species range maps (SRMs) are essential tools for research and policy-making in ecology, conservation, and environmental management. However, traditional SRMs rely on the availability of environmental covariates and high-quality species location observation data, both of which can be challenging to obtain due to geographic inaccessibility and resource constraints. We propose a novel approach combining millions of citizen science species observations with textual descriptions from Wikipedia, covering habitat preferences and range descriptions for tens of thousands of species. Our framework maps locations, species, and text descriptions into a common space, facilitating the learning of rich spatial covariates at a global scale and enabling zero-shot range estimation from textual descriptions. Evaluated on held-out species, our zero-shot SRMs significantly outperform baselines and match the performance of SRMs obtained using tens of observations. Our approach also acts as a strong prior when combined with observational data, resulting in more accurate range estimation with less data. We present extensive quantitative and qualitative analyses of the learned representations in the context of range estimation and other spatial tasks, demonstrating the effectiveness of our approach.
Authors: Sepideh Vafaie, Deepak Bal, Michael A. S. Thorne, Eric Forgoston
Abstract: For years, a main focus of ecological research has been to better understand the complex dynamical interactions between species which comprise food webs. Using the connectance properties of a widely explored synthetic food web called the cascade model, we explore the behavior of dynamics on Lotka-Volterra ecological systems. We show how trophic efficiency, a staple assumption in mathematical ecology, produces systems which are not persistent. With clustering analysis we show how straightforward inequalities of the summed values of the birth, death, self-regulation and interaction strengths provide insight into which food webs are more enduring or stable. Through these simplified summed values, we develop a random forest model and a neural network model, both of which are able to predict the number of extinctions that would occur without the need to simulate the dynamics. To conclude, we highlight the variable that plays the dominant role in determining the order in which species go extinct.
Authors: Haozhen Zhang, Tao Feng, Jiaxuan You
Abstract: Retrieval-augmented generation (RAG) has revitalized Large Language Models (LLMs) by injecting non-parametric factual knowledge. Compared with long-context LLMs, RAG is considered an effective summarization tool in a more concise and lightweight manner, which can interact with LLMs multiple times using diverse queries to get comprehensive responses. However, the LLM-generated historical responses, which contain potentially insightful information, are largely neglected and discarded by existing approaches, leading to suboptimal results. In this paper, we propose \textit{graph of records} (\textbf{GoR}), which leverages historical responses generated by LLMs to enhance RAG for long-context global summarization. Inspired by the \textit{retrieve-then-generate} paradigm of RAG, we construct a graph by establishing an edge between the retrieved text chunks and the corresponding LLM-generated response. To further uncover the intricate correlations between them, GoR further features a \textit{graph neural network} and an elaborately designed \textit{BERTScore}-based objective for self-supervised model training, enabling seamless supervision signal backpropagation between reference summaries and node embeddings. We comprehensively compare GoR with 12 baselines across four long-context summarization datasets, and the results indicate that our proposed method reaches the best performance e.g., 15\%, 8\%, and 19\% improvement over retrievers w.r.t. Rouge-L, Rouge-1, and Rouge-2 on the WCEP dataset). Extensive experiments further demonstrate the effectiveness of GoR. Code is available at https://github.com/ulab-uiuc/GoR
Authors: Fangru Lin, Shaoguang Mao, Emanuele La Malfa, Valentin Hofmann, Adrian de Wynter, Jing Yao, Si-Qing Chen, Michael Wooldridge, Furu Wei
Abstract: Language is not monolithic. While many benchmarks are used as proxies to systematically estimate Large Language Models' (LLM) performance in real-life tasks, they tend to ignore the nuances of within-language variation and thus fail to model the experience of speakers of minority dialects. Focusing on African American Vernacular English (AAVE), we present the first study on LLMs' fairness and robustness to a dialect in canonical reasoning tasks (algorithm, math, logic, and comprehensive reasoning). We hire AAVE speakers, including experts with computer science backgrounds, to rewrite seven popular benchmarks, such as HumanEval and GSM8K. The result of this effort is ReDial, a dialectal benchmark comprising $1.2K+$ parallel query pairs in Standardized English and AAVE. We use ReDial to evaluate state-of-the-art LLMs, including GPT-4o/4/3.5-turbo, LLaMA-3.1/3, Mistral, and Phi-3. We find that, compared to Standardized English, almost all of these widely used models show significant brittleness and unfairness to queries in AAVE. Furthermore, AAVE queries can degrade performance more substantially than misspelled texts in Standardized English, even when LLMs are more familiar with the AAVE queries. Finally, asking models to rephrase questions in Standardized English does not close the performance gap but generally introduces higher costs. Overall, our findings indicate that LLMs provide unfair service to dialect users in complex reasoning tasks. Code can be found at https://github.com/fangru-lin/redial_dialect_robustness_fairness.git.
URLs: https://github.com/fangru-lin/redial_dialect_robustness_fairness.git.
Authors: Benjamin Towle, Ke Zhou
Abstract: AI-mediated communication enables users to communicate more quickly and efficiently. Various systems have been proposed such as smart reply and AI-assisted writing. Yet, the heterogeneity of the forms of inputs and architectures often renders it challenging to combine insights from user behaviour in one system to improve performance in another. In this work, we consider the case where the user does not select any of the suggested replies from a smart reply system, and how this can be used as one-shot implicit negative feedback to enhance the accuracy of an AI writing model. We introduce Nifty, an approach that uses classifier guidance to controllably integrate implicit user feedback into the text generation process. Empirically, we find up to 34% improvement in Rouge-L, 89% improvement in generating the correct intent, and an 86% win-rate according to human evaluators compared to a vanilla AI writing system on the MultiWOZ and Schema-Guided Dialog datasets.
Authors: Efimia Panagiotaki, Daniele De Martini, Lars Kunze, Petar Veli\v{c}kovi\'c
Abstract: This study explores the intersection of neural networks and classical robotics algorithms through the Neural Algorithmic Reasoning (NAR) framework, allowing to train neural networks to effectively reason like classical robotics algorithms by learning to execute them. Algorithms are integral to robotics and safety-critical applications due to their predictable and consistent performance through logical and mathematical principles. In contrast, while neural networks are highly adaptable, handling complex, high-dimensional data and generalising across tasks, they often lack interpretability and transparency in their internal computations. We propose a Graph Neural Network (GNN)-based learning framework, NAR-*ICP, which learns the intermediate algorithmic steps of classical ICP-based pointcloud registration algorithms, and extend the CLRS Algorithmic Reasoning Benchmark with classical robotics perception algorithms. We evaluate our approach across diverse datasets, from real-world to synthetic, demonstrating its flexibility in handling complex and noisy inputs, along with its potential to be used as part of a larger learning system. Our results indicate that our method achieves superior performance across all benchmarks and datasets, consistently surpassing even the algorithms it has been trained on, further demonstrating its ability to generalise beyond the capabilities of traditional algorithms.
Authors: Yuri Gardinazzi, Giada Panerai, Karthik Viswanathan, Alessio Ansuini, Alberto Cazzaniga, Matteo Biagetti
Abstract: Understanding the decision-making processes of large language models (LLMs) is critical given their widespread applications. Towards this goal, describing the topological and geometrical properties of internal representations has recently provided valuable insights. For a more comprehensive characterization of these inherently complex spaces, we present a novel framework based on zigzag persistence, a method in topological data analysis (TDA) well-suited for describing data undergoing dynamic transformations across layers. Within this framework, we introduce persistence similarity, a new metric that quantifies the persistence and transformation of topological features such as $p$-cycles throughout the model layers. Unlike traditional similarity measures, our approach captures the entire evolutionary trajectory of these features, providing deeper insights into the internal workings of LLMs. As a practical application, we leverage persistence similarity to identify and prune redundant layers, demonstrating comparable performance to state-of-the-art methods across several benchmark datasets. Additionally, our analysis reveals consistent topological behaviors across various models and hyperparameter settings, suggesting a universal structure in LLM internal representations.
Authors: Liang Tao, Yixin Xie, Jeffrey D Deng, Hui Shen, Hong-Wen Deng, Weihua Zhou, Chen Zhao
Abstract: Alzheimer's disease (AD) is a chronic neurodegenerative disorder and the leading cause of dementia, significantly impacting cost, mortality, and burden worldwide. The advent of high-throughput omics technologies, such as genomics, transcriptomics, proteomics, and epigenomics, has revolutionized the molecular understanding of AD. Conventional AI approaches typically require the completion of all omics data at the outset to achieve optimal AD diagnosis, which are inefficient and may be unnecessary. To reduce the clinical cost and improve the accuracy of AD diagnosis using multi-omics data, we propose a novel staged graph convolutional network with uncertainty quantification (SGUQ). SGUQ begins with mRNA and progressively incorporates DNA methylation and miRNA data only when necessary, reducing overall costs and exposure to harmful tests. Experimental results indicate that 46.23% of the samples can be reliably predicted using only single-modal omics data (mRNA), while an additional 16.04% of the samples can achieve reliable predictions when combining two omics data types (mRNA + DNA methylation). In addition, the proposed staged SGUQ achieved an accuracy of 0.858 on ROSMAP dataset, which outperformed existing methods significantly. The proposed SGUQ can not only be applied to AD diagnosis using multi-omics data but also has the potential for clinical decision-making using multi-viewed data. Our implementation is publicly available at https://github.com/chenzhao2023/multiomicsuncertainty.
URLs: https://github.com/chenzhao2023/multiomicsuncertainty.
Authors: Charles C. Margossian, Lawrence K. Saul
Abstract: Given an intractable target density $p$, variational inference (VI) attempts to find the best approximation $q$ from a tractable family $Q$. This is typically done by minimizing the exclusive Kullback-Leibler divergence, $\text{KL}(q||p)$. In practice, $Q$ is not rich enough to contain $p$, and the approximation is misspecified even when it is a unique global minimizer of $\text{KL}(q||p)$. In this paper, we analyze the robustness of VI to these misspecifications when $p$ exhibits certain symmetries and $Q$ is a location-scale family that shares these symmetries. We prove strong guarantees for VI not only under mild regularity conditions but also in the face of severe misspecifications. Namely, we show that (i) VI recovers the mean of $p$ when $p$ exhibits an \textit{even} symmetry, and (ii) it recovers the correlation matrix of $p$ when in addition~$p$ exhibits an \textit{elliptical} symmetry. These guarantees hold for the mean even when $q$ is factorized and $p$ is not, and for the correlation matrix even when~$q$ and~$p$ behave differently in their tails. We analyze various regimes of Bayesian inference where these symmetries are useful idealizations, and we also investigate experimentally how VI behaves in their absence.
Authors: Jaesung Huh, Andrew Zisserman
Abstract: This paper presents an improved framework for character-aware audio-visual subtitling in TV shows. Our approach integrates speech recognition, speaker diarisation, and character recognition, utilising both audio and visual cues. This holistic solution addresses what is said, when it's said, and who is speaking, providing a more comprehensive and accurate character-aware subtitling for TV shows. Our approach brings improvements on two fronts: first, we show that audio-visual synchronisation can be used to pick out the talking face amongst others present in a video clip, and assign an identity to the corresponding speech segment. This audio-visual approach improves recognition accuracy and yield over current methods. Second, we show that the speaker of short segments can be determined by using the temporal context of the dialogue within a scene. We propose an approach using local voice embeddings of the audio, and large language model reasoning on the text transcription. This overcomes a limitation of existing methods that they are unable to accurately assign speakers to short temporal segments. We validate the method on a dataset with 12 TV shows, demonstrating superior performance in speaker diarisation and character recognition accuracy compared to existing approaches. Project page : https://www.robots.ox.ac.uk/~vgg/research/llr-context/
Authors: Yuqi Su, Xiaolei Fang
Abstract: Industrial prognostics aims to develop data-driven methods that leverage high-dimensional degradation signals from assets to predict their failure times. The success of these models largely depends on the availability of substantial historical data for training. However, in practice, individual organizations often lack sufficient data to independently train reliable prognostic models, and privacy concerns prevent data sharing between organizations for collaborative model training. To overcome these challenges, this article proposes a statistical learning-based federated model that enables multiple organizations to jointly train a prognostic model while keeping their data local and secure. The proposed approach involves two key stages: federated dimension reduction and federated (log)-location-scale regression. In the first stage, we develop a federated randomized singular value decomposition algorithm for multivariate functional principal component analysis, which efficiently reduces the dimensionality of degradation signals while maintaining data privacy. The second stage proposes a federated parameter estimation algorithm for (log)-location-scale regression, allowing organizations to collaboratively estimate failure time distributions without sharing raw data. The proposed approach addresses the limitations of existing federated prognostic methods by using statistical learning techniques that perform well with smaller datasets and provide comprehensive failure time distributions. The effectiveness and practicality of the proposed model are validated using simulated data and a dataset from the NASA repository.
Authors: Elizabeth Teng, Ugur Demir, Zoheyr Doctor, Philipp M. Srivastava, Shamal Lalvani, Vicky Kalogera, Aggelos Katsaggelos, Jeff J. Andrews, Simone S. Bavera, Max M. Briel, Seth Gossage, Konstantinos Kovlakas, Matthias U. Kruckow, Kyle Akira Rocha, Meng Sun, Zepei Xing, Emmanouil Zapartas
Abstract: Knowledge about the internal physical structure of stars is crucial to understanding their evolution. The novel binary population synthesis code POSYDON includes a module for interpolating the stellar and binary properties of any system at the end of binary MESA evolution based on a pre-computed set of models. In this work, we present a new emulation method for predicting stellar profiles, i.e., the internal stellar structure along the radial axis, using machine learning techniques. We use principal component analysis for dimensionality reduction and fully-connected feed-forward neural networks for making predictions. We find accuracy to be comparable to that of nearest neighbor approximation, with a strong advantage in terms of memory and storage efficiency. By delivering more information about the evolution of stellar internal structure, these emulators will enable faster simulations of higher physical fidelity with large-scale simulations of binary star population synthesis possible with POSYDON and other population synthesis codes.
Authors: Luis Chuquimarca, Boris Vintimilla, Sergio Velastin
Abstract: This study presents an investigation into the utilization of a Multi-Input architecture for the classification of fruits (apples and mangoes) into healthy and defective states, employing both RGB and silhouette images. The primary aim is to enhance the accuracy of CNN models. The methodology encompasses image acquisition, preprocessing of datasets, training, and evaluation of two CNN models: MobileNetV2 and VGG16. Results reveal that the inclusion of silhouette images alongside the Multi-Input architecture yields models with superior performance compared to using only RGB images for fruit classification, whether healthy or defective. Specifically, optimal results were achieved using the MobileNetV2 model, achieving 100\% accuracy. This finding suggests the efficacy of this combined methodology in improving the precise classification of healthy or defective fruits, which could have significant implications for applications related to external quality inspection of fruits.
Authors: Chad Brown
Abstract: This paper establishes statistical properties of deep neural network (DNN) estimators under dependent data. Two general results for nonparametric sieve estimators directly applicable to DNNs estimators are given. The first establishes rates for convergence in probability under nonstationary data. The second provides non-asymptotic probability bounds on $\mathcal{L}^{2}$-errors under stationary $\beta$-mixing data. I apply these results to DNN estimators in both regression and classification contexts imposing only a standard H\"older smoothness assumption. These results are then used to demonstrate how asymptotic inference can be conducted on the finite dimensional parameter of a partially linear regression model after first-stage DNN estimation of infinite dimensional parameters. The DNN architectures considered are common in applications, featuring fully connected feedforward networks with any continuous piecewise linear activation function, unbounded weights, and a width and depth that grows with sample size. The framework provided also offers potential for research into other DNN architectures and time-series applications.
Authors: Yiping Lu, Daozhe Lin, Qiang Du
Abstract: In this paper, we establish a novel connection between the metric entropy growth and the embeddability of function spaces into reproducing kernel Hilbert/Banach spaces. Metric entropy characterizes the information complexity of function spaces and has implications for their approximability and learnability. Classical results show that embedding a function space into a reproducing kernel Hilbert space (RKHS) implies a bound on its metric entropy growth. Surprisingly, we prove a \textbf{converse}: a bound on the metric entropy growth of a function space allows its embedding to a $\mathcal{L}_p-$type Reproducing Kernel Banach Space (RKBS). This shows that the $\mathcal{L}_p-$type RKBS provides a broad modeling framework for learnable function classes with controlled metric entropies. Our results shed new light on the power and limitations of kernel methods for learning complex function spaces.
Authors: Kangning Cui, Wei Tang, Rongkun Zhu, Manqi Wang, Gregory D. Larsen, Victor P. Pauca, Sarra Alqahtani, Fan Yang, David Segurado, Paul Fine, Jordan Karubian, Raymond H. Chan, Robert J. Plemmons, Jean-Michel Morel, Miles R. Silman
Abstract: Understanding the spatial distribution of palms within tropical forests is essential for effective ecological monitoring, conservation strategies, and the sustainable integration of natural forest products into local and global supply chains. However, the analysis of remotely sensed data in these environments faces significant challenges, such as overlapping palm and tree crowns, uneven shading across the canopy surface, and the heterogeneous nature of the forest landscapes, which often affect the performance of palm detection and segmentation algorithms. To overcome these issues, we introduce PalmDSNet, a deep learning framework for real-time detection, segmentation, and counting of canopy palms. Additionally, we employ a bimodal reproduction algorithm that simulates palm spatial propagation to further enhance the understanding of these point patterns using PalmDSNet's results. We used UAV-captured imagery to create orthomosaics from 21 sites across western Ecuadorian tropical forests, covering a gradient from the everwet Choc\'o forests near Colombia to the drier forests of southwestern Ecuador. Expert annotations were used to create a comprehensive dataset, including 7,356 bounding boxes on image patches and 7,603 palm centers across five orthomosaics, encompassing a total area of 449 hectares. By combining PalmDSNet with the bimodal reproduction algorithm, which optimizes parameters for both local and global spatial variability, we effectively simulate the spatial distribution of palms in diverse and dense tropical environments, validating its utility for advanced applications in tropical forest monitoring and remote sensing analysis.
Authors: Yaxuan Wang, Jiaheng Wei, Chris Yuhao Liu, Jinlong Pang, Quan Liu, Ankit Parag Shah, Yujia Bao, Yang Liu, Wei Wei
Abstract: Unlearning in Large Language Models (LLMs) is essential for ensuring ethical and responsible AI use, especially in addressing privacy leak, bias, safety, and evolving regulations. Existing approaches to LLM unlearning often rely on retain data or a reference LLM, yet they struggle to adequately balance unlearning performance with overall model utility. This challenge arises because leveraging explicit retain data or implicit knowledge of retain data from a reference LLM to fine-tune the model tends to blur the boundaries between the forgotten and retain data, as different queries often elicit similar responses. In this work, we propose eliminating the need to retain data or the reference LLM for response calibration in LLM unlearning. Recognizing that directly applying gradient ascent on the forget data often leads to optimization instability and poor performance, our method guides the LLM on what not to respond to, and importantly, how to respond, based on the forget data. Hence, we introduce Forget data only Loss AjustmenT (FLAT), a "flat" loss adjustment approach which addresses these issues by maximizing f-divergence between the available template answer and the forget answer only w.r.t. the forget data. The variational form of the defined f-divergence theoretically provides a way of loss adjustment by assigning different importance weights for the learning w.r.t. template responses and the forgetting of responses subject to unlearning. Empirical results demonstrate that our approach not only achieves superior unlearning performance compared to existing methods but also minimizes the impact on the model's retained capabilities, ensuring high utility across diverse tasks, including copyrighted content unlearning on Harry Potter dataset and MUSE Benchmark, and entity unlearning on the TOFU dataset.
Authors: Andrew Levy, Alessandro Allievi, George Konidaris
Abstract: Empowerment has the potential to help agents learn large skillsets, but is not yet a scalable solution for training general-purpose agents. Recent empowerment methods learn diverse skillsets by maximizing the mutual information between skills and states; however, these approaches require a model of the transition dynamics, which can be challenging to learn in realistic settings with high-dimensional and stochastic observations. We present Latent-Predictive Empowerment (LPE), an algorithm that can compute empowerment in a more practical manner. LPE learns large skillsets by maximizing an objective that is a principled replacement for the mutual information between skills and states and that only requires a simpler latent-predictive model rather than a full simulator of the environment. We show empirically in a variety of settings--including ones with high-dimensional observations and highly stochastic transition dynamics--that our empowerment objective (i) learns similar-sized skillsets as the leading empowerment algorithm that assumes access to a model of the transition dynamics and (ii) outperforms other model-based approaches to empowerment.
Authors: Nick McGreivy
Abstract: Differentiable programming allows for derivatives of functions implemented via computer code to be calculated automatically. These derivatives are calculated using automatic differentiation (AD). This thesis explores two applications of differentiable programming to computational plasma physics. First, we consider how differentiable programming can be used to simplify and improve stellarator optimization. We introduce a stellarator coil design code (FOCUSADD) that uses gradient-based optimization to produce stellarator coils with finite build. Because we use reverse mode AD, which can compute gradients of scalar functions with the same computational complexity as the function, FOCUSADD is simple, flexible, and efficient. We then discuss two additional applications of AD in stellarator optimization. Second, we explore how machine learning (ML) can be used to improve or replace the numerical methods used to solve partial differential equations (PDEs), focusing on time-dependent PDEs in fluid mechanics relevant to plasma physics. Differentiable programming allows neural networks and other techniques from ML to be embedded within numerical methods. This is a promising, but relatively new, research area. We focus on two basic questions. First, can we design ML-based PDE solvers that have the same guarantees of conservation, stability, and positivity that standard numerical methods do? The answer is yes; we introduce error-correcting algorithms that preserve invariants of time-dependent PDEs. Second, which types of ML-based solvers work best at solving PDEs? We perform a systematic review of the scientific literature on solving PDEs with ML. Unfortunately we discover two issues, weak baselines and reporting biases, that affect the interpretation reproducibility of a significant majority of published research. We conclude that using ML to solve PDEs is not as promising as we initially believed.
Authors: Sheng Yan, Cunhang fan, Hongyu Zhang, Xiaoke Yang, Jianhua Tao, Zhao Lv
Abstract: At a cocktail party, humans exhibit an impressive ability to direct their attention. The auditory attention detection (AAD) approach seeks to identify the attended speaker by analyzing brain signals, such as EEG signals. However, current AAD algorithms overlook the spatial distribution information within EEG signals and lack the ability to capture long-range latent dependencies, limiting the model's ability to decode brain activity. To address these issues, this paper proposes a dual attention refinement network with spatiotemporal construction for AAD, named DARNet, which consists of the spatiotemporal construction module, dual attention refinement module, and feature fusion \& classifier module. Specifically, the spatiotemporal construction module aims to construct more expressive spatiotemporal feature representations, by capturing the spatial distribution characteristics of EEG signals. The dual attention refinement module aims to extract different levels of temporal patterns in EEG signals and enhance the model's ability to capture long-range latent dependencies. The feature fusion \& classifier module aims to aggregate temporal patterns and dependencies from different levels and obtain the final classification results. The experimental results indicate that compared to the state-of-the-art models, DARNet achieves an average classification accuracy improvement of 5.9\% for 0.1s, 4.6\% for 1s, and 3.9\% for 2s on the DTU dataset. While maintaining excellent classification performance, DARNet significantly reduces the number of required parameters. Compared to the state-of-the-art models, DARNet reduces the parameter count by 91\%. Code is available at: https://github.com/fchest/DARNet.git.
Authors: Suma Anand, Kaiwen Xu, Colm O'Dushlaine, Sumit Mukherjee
Abstract: Magnetic Resonance Imaging (MRI) is the gold standard for measuring fat and iron content non-invasively in the body via measures known as Proton Density Fat Fraction (PDFF) and $R_2^*$, respectively. However, conventional PDFF and $R_2^*$ quantification methods operate on MR images voxel-wise and require at least three measurements to estimate three quantities: water, fat, and $R_2^*$. Alternatively, the two-point Dixon MRI protocol is widely used and fast because it acquires only two measurements; however, these cannot be used to estimate three quantities voxel-wise. Leveraging the fact that neighboring voxels have similar values, we propose using a generative machine learning approach to learn PDFF and $R_2^*$ from Dixon MRI. We use paired Dixon-IDEAL data from UK Biobank in the liver and a Pix2Pix conditional GAN to demonstrate the first large-scale $R_2^*$ imputation from two-point Dixon MRIs. Using our proposed approach, we synthesize PDFF and $R_2^*$ maps that show significantly greater correlation with ground-truth than conventional voxel-wise baselines.
Authors: Zhifei Xie, Changqiao Wu
Abstract: GPT4o, an all-encompassing model, represents a milestone in the development of multi-modal large models. It can understand visual, auditory, and textual modalities, directly output audio, and support flexible duplex interaction. However, its technical framework is not open-sourced. Models from the open-source community often achieve some functionalities of GPT4o, such as visual understanding and voice dialogue. Nevertheless, training a unified model that incorporates all modalities is challenging due to the complexities of multi-modal data, intricate model architectures, and training processes. In this paper, we introduce Mini-Omni2, a visual-audio assistant capable of providing real-time, end-to-end voice responses to user video and voice queries, while also incorporating auditory capabilities. By integrating pretrained visual and auditory encoders, Mini-Omni2 maintains strong performance in individual modalities. We propose a three-stage training process to align modalities, allowing the language model to handle multi-modal inputs and outputs after training on a limited dataset. For interaction, we introduce a semantic-based interruption mechanism, enabling more flexible dialogues with users. All modeling approaches and data construction methods will be open-sourced. To the best of our knowledge, Mini-Omni2 is one of the models closest to GPT4o in functionality, and we hope it can offer valuable insights for subsequent research.
Authors: Tong Ding, Wanhua Li, Zhongqi Miao, Hanspeter Pfister
Abstract: Prompt learning has proven effective in adapting vision language models for downstream tasks. However, existing methods usually append learnable prompt tokens solely with the category names to obtain textual features, which fails to fully leverage the rich context indicated in the category name. To address this issue, we propose the Tree of Attributes Prompt learning (TAP), which first instructs LLMs to generate a tree of attributes with a "concept - attribute - description" structure for each category, and then learn the hierarchy with vision and text prompt tokens. Unlike existing methods that merely augment category names with a set of unstructured descriptions, our approach essentially distills structured knowledge graphs associated with class names from LLMs. Furthermore, our approach introduces text and vision prompts designed to explicitly learn the corresponding visual attributes, effectively serving as domain experts. Additionally, the general and diverse descriptions generated based on the class names may be wrong or absent in the specific given images. To address this misalignment, we further introduce a vision-conditional pooling module to extract instance-specific text features. Extensive experimental results demonstrate that our approach outperforms state-of-the-art methods on the zero-shot base-to-novel generalization, cross-dataset transfer, as well as few-shot classification across 11 diverse datasets.
Authors: Zhengyang Yu, Zhaoyuan Yang, Jing Zhang
Abstract: Recent text-to-image personalization methods have shown great promise in teaching a diffusion model user-specified concepts given a few images for reusing the acquired concepts in a novel context. With massive efforts being dedicated to personalized generation, a promising extension is personalized editing, namely to edit an image using personalized concepts, which can provide a more precise guidance signal than traditional textual guidance. To address this, a straightforward solution is to incorporate a personalized diffusion model with a text-driven editing framework. However, such a solution often shows unsatisfactory editability on the source image. To address this, we propose DreamSteerer, a plug-in method for augmenting existing T2I personalization methods. Specifically, we enhance the source image conditioned editability of a personalized diffusion model via a novel Editability Driven Score Distillation (EDSD) objective. Moreover, we identify a mode trapping issue with EDSD, and propose a mode shifting regularization with spatial feature guided sampling to avoid such an issue. We further employ two key modifications to the Delta Denoising Score framework that enable high-fidelity local editing with personalized concepts. Extensive experiments validate that DreamSteerer can significantly improve the editability of several T2I personalization baselines while being computationally efficient.
Authors: Pranav Gupta, Rishabh Rengarajan, Viren Bankapur, Vedansh Mannem, Lakshit Ahuja, Surya Vijay, Kevin Wang
Abstract: Combining LiDAR and Camera-view data has become a common approach for 3D Object Detection. However, previous approaches combine the two input streams at a point-level, throwing away semantic information derived from camera features. In this paper we propose Cross-View Center Point-Fusion, a state-of-the-art model to perform 3D object detection by combining camera and LiDAR-derived features in the BEV space to preserve semantic density from the camera stream while incorporating spacial data from the LiDAR stream. Our architecture utilizes aspects from previously established algorithms, Cross-View Transformers and CenterPoint, and runs their backbones in parallel, allowing efficient computation for real-time processing and application. In this paper we find that while an implicitly calculated depth-estimate may be sufficiently accurate in a 2D map-view representation, explicitly calculated geometric and spacial information is needed for precise bounding box prediction in the 3D world-view space.
Authors: Pedram Akbarian, Huy Nguyen, Xing Han, Nhat Ho
Abstract: Mixture of Experts (MoE) models are highly effective in scaling model capacity while preserving computational efficiency, with the gating network, or router, playing a central role by directing inputs to the appropriate experts. In this paper, we establish a novel connection between MoE frameworks and attention mechanisms, demonstrating how quadratic gating can serve as a more expressive and efficient alternative. Motivated by this insight, we explore the implementation of quadratic gating within MoE models, identifying a connection between the self-attention mechanism and the quadratic gating. We conduct a comprehensive theoretical analysis of the quadratic softmax gating MoE framework, showing improved sample efficiency in expert and parameter estimation. Our analysis provides key insights into optimal designs for quadratic gating and expert functions, further elucidating the principles behind widely used attention mechanisms. Through extensive evaluations, we demonstrate that the quadratic gating MoE outperforms the traditional linear gating MoE. Moreover, our theoretical insights have guided the development of a novel attention mechanism, which we validated through extensive experiments. The results demonstrate its favorable performance over conventional models across various tasks.
Authors: Jiaxian Yan, Zaixi Zhang, Jintao Zhu, Kai Zhang, Jianfeng Pei, Qi Liu
Abstract: Molecular docking, a technique for predicting ligand binding poses, is crucial in structure-based drug design for understanding protein-ligand interactions. Recent advancements in docking methods, particularly those leveraging geometric deep learning (GDL), have demonstrated significant efficiency and accuracy advantages over traditional sampling methods. Despite these advancements, current methods are often tailored for specific docking settings, and limitations such as the neglect of protein side-chain structures, difficulties in handling large binding pockets, and challenges in predicting physically valid structures exist. To accommodate various docking settings and achieve accurate, efficient, and physically reliable docking, we propose a novel two-stage docking framework, DeltaDock, consisting of pocket prediction and site-specific docking. We innovatively reframe the pocket prediction task as a pocket-ligand alignment problem rather than direct prediction in the first stage. Then we follow a bi-level coarse-to-fine iterative refinement process to perform site-specific docking. Comprehensive experiments demonstrate the superior performance of DeltaDock. Notably, in the blind docking setting, DeltaDock achieves a 31\% relative improvement over the docking success rate compared with the previous state-of-the-art GDL model. With the consideration of physical validity, this improvement increases to about 300\%.
Authors: Thomas T. Zhang, Bruce D. Lee, Ingvar Ziemann, George J. Pappas, Nikolai Matni
Abstract: A driving force behind the diverse applicability of modern machine learning is the ability to extract meaningful features across many sources. However, many practical domains involve data that are non-identically distributed across sources, and statistically dependent within its source, violating vital assumptions in existing theoretical studies. Toward addressing these issues, we establish statistical guarantees for learning general $\textit{nonlinear}$ representations from multiple data sources that admit different input distributions and possibly dependent data. Specifically, we study the sample-complexity of learning $T+1$ functions $f_\star^{(t)} \circ g_\star$ from a function class $\mathcal F \times \mathcal G$, where $f_\star^{(t)}$ are task specific linear functions and $g_\star$ is a shared nonlinear representation. A representation $\hat g$ is estimated using $N$ samples from each of $T$ source tasks, and a fine-tuning function $\hat f^{(0)}$ is fit using $N'$ samples from a target task passed through $\hat g$. We show that when $N \gtrsim C_{\mathrm{dep}} (\mathrm{dim}(\mathcal F) + \mathrm{C}(\mathcal G)/T)$, the excess risk of $\hat f^{(0)} \circ \hat g$ on the target task decays as $\nu_{\mathrm{div}} \big(\frac{\mathrm{dim}(\mathcal F)}{N'} + \frac{\mathrm{C}(\mathcal G)}{N T} \big)$, where $C_{\mathrm{dep}}$ denotes the effect of data dependency, $\nu_{\mathrm{div}}$ denotes an (estimatable) measure of $\textit{task-diversity}$ between the source and target tasks, and $\mathrm C(\mathcal G)$ denotes the complexity of the representation class $\mathcal G$. In particular, our analysis reveals: as the number of tasks $T$ increases, both the sample requirement and risk bound converge to that of $r$-dimensional regression as if $g_\star$ had been given, and the effect of dependency only enters the sample requirement, leaving the risk bound matching the iid setting.
Authors: Zhongye Liu, Hongbin Liu, Yuepeng Hu, Zedian Shao, Neil Zhenqiang Gong
Abstract: Visual hallucination (VH) occurs when a multimodal large language model (MLLM) generates responses with incorrect visual details for prompts. Existing methods for generating VH test cases primarily rely on human annotations, typically in the form of triples: (image, question, answer). In this paper, we introduce VHExpansion, the first automated method for expanding VH test cases for MLLMs. Given an initial VH test case, VHExpansion automatically expands it by perturbing the question and answer through negation as well as modifying the image using both common and adversarial perturbations. Additionally, we propose a new evaluation metric, symmetric accuracy, which measures the proportion of correctly answered VH test-case pairs. Each pair consists of a test case and its negated counterpart. Our theoretical analysis shows that symmetric accuracy is an unbiased evaluation metric that remains unaffected by the imbalance of VH testing cases with varying answers when an MLLM is randomly guessing the answers, whereas traditional accuracy is prone to such imbalance. We apply VHExpansion to expand three VH datasets annotated manually and use these expanded datasets to benchmark seven MLLMs. Our evaluation shows that VHExpansion effectively identifies more VH test cases. Moreover, symmetric accuracy, being unbiased, leads to different conclusions about the vulnerability of MLLMs to VH compared to traditional accuracy metric. Finally, we show that fine-tuning MLLMs on the expanded VH dataset generated by VHExpansion mitigates VH more effectively than fine-tuning on the original, manually annotated dataset. Our code is available at: https://github.com/lycheeefish/VHExpansion.
Authors: Rajesh Mangannavar, Gopalakrishnan Srinivasaraghavan
Abstract: Among the many variants of RL, an important class of problems is where the state and action spaces are continuous -- autonomous robots, autonomous vehicles, optimal control are all examples of such problems that can lend themselves naturally to reinforcement based algorithms, and have continuous state and action spaces. In this paper, we introduce a prioritized form of a combination of state-of-the-art approaches such as Deep Q-learning (DQN) and Deep Deterministic Policy Gradient (DDPG) to outperform the earlier results for continuous state and action space problems. Our experiments also involve the use of parameter noise during training resulting in more robust deep RL models outperforming the earlier results significantly. We believe these results are a valuable addition for continuous state and action space problems.
Authors: Carl Henrik Ek, Oisin Kim, Challenger Mishra
Abstract: Motivated by recent progress in the problem of numerical K\"ahler metrics, we survey machine learning techniques in this area, discussing both advantages and drawbacks. We then revisit the algebraic ansatz pioneered by Donaldson. Inspired by his work, we present a novel approach to obtaining Ricci-flat approximations to K\"ahler metrics, applying machine learning within a `principled' framework. In particular, we use gradient descent on the Grassmannian manifold to identify an efficient subspace of sections for calculation of the metric. We combine this approach with both Donaldson's algorithm and learning on the $h$-matrix itself (the latter method being equivalent to gradient descent on the fibre bundle of Hermitian metrics on the tautological bundle over the Grassmannian). We implement our methods on the Dwork family of threefolds, commenting on the behaviour at different points in moduli space. In particular, we observe the emergence of nontrivial local minima as the moduli parameter is increased.
Authors: Matheus Farias, H. T. Kung
Abstract: We introduce $\textit{sorted weight sectioning}$ (SWS): a weight allocation algorithm that places sorted deep neural network (DNN) weight sections on bit-sliced compute-in-memory (CIM) crossbars to reduce analog-to-digital converter (ADC) energy consumption. Data conversions are the most energy-intensive process in crossbar operation. SWS effectively reduces this cost leveraging (1) small weights and (2) zero weights (weight sparsity). DNN weights follow bell-shaped distributions, with most weights near zero. Using SWS, we only need low-order crossbar columns for sections with low-magnitude weights. This reduces the quantity and resolution of ADCs used, exponentially decreasing ADC energy costs without significantly degrading DNN accuracy. Unstructured sparsification further sharpens the weight distribution with small accuracy loss. However, it presents challenges in hardware tracking of zeros: we cannot switch zero rows to other layer weights in unsorted crossbars without index matching. SWS efficiently addresses unstructured sparse models using offline remapping of zeros into earlier sections, which reveals full sparsity potential and maximizes energy efficiency. Our method reduces ADC energy use by 89.5% on unstructured sparse BERT models. Overall, this paper introduces a novel algorithm to promote energy-efficient CIM crossbars for unstructured sparse DNN workloads.
Authors: Shuo Li, Tao Ji, Xiaoran Fan, Linsheng Lu, Leyi Yang, Yuming Yang, Zhiheng Xi, Rui Zheng, Yuran Wang, Xiaohui Zhao, Tao Gui, Qi Zhang, Xuanjing Huang
Abstract: In the study of LLMs, sycophancy represents a prevalent hallucination that poses significant challenges to these models. Specifically, LLMs often fail to adhere to original correct responses, instead blindly agreeing with users' opinions, even when those opinions are incorrect or malicious. However, research on sycophancy in visual language models (VLMs) has been scarce. In this work, we extend the exploration of sycophancy from LLMs to VLMs, introducing the MM-SY benchmark to evaluate this phenomenon. We present evaluation results from multiple representative models, addressing the gap in sycophancy research for VLMs. To mitigate sycophancy, we propose a synthetic dataset for training and employ methods based on prompts, supervised fine-tuning, and DPO. Our experiments demonstrate that these methods effectively alleviate sycophancy in VLMs. Additionally, we probe VLMs to assess the semantic impact of sycophancy and analyze the attention distribution of visual tokens. Our findings indicate that the ability to prevent sycophancy is predominantly observed in higher layers of the model. The lack of attention to image knowledge in these higher layers may contribute to sycophancy, and enhancing image attention at high layers proves beneficial in mitigating this issue.
Authors: Gaoyang Pang, Wanchun Liu, Dusit Niyato, Branka Vucetic, Yonghui Li
Abstract: Wireless Networked Control Systems (WNCSs) are essential to Industry 4.0, enabling flexible control in applications, such as drone swarms and autonomous robots. The interdependence between communication and control requires integrated design, but traditional methods treat them separately, leading to inefficiencies. Current codesign approaches often rely on simplified models, focusing on single-loop or independent multi-loop systems. However, large-scale WNCSs face unique challenges, including coupled control loops, time-correlated wireless channels, trade-offs between sensing and control transmissions, and significant computational complexity. To address these challenges, we propose a practical WNCS model that captures correlated dynamics among multiple control loops with spatially distributed sensors and actuators sharing limited wireless resources over multi-state Markov block-fading channels. We formulate the codesign problem as a sequential decision-making task that jointly optimizes scheduling and control inputs across estimation, control, and communication domains. To solve this problem, we develop a Deep Reinforcement Learning (DRL) algorithm that efficiently handles the hybrid action space, captures communication-control correlations, and ensures robust training despite sparse cross-domain variables and floating control inputs. Extensive simulations show that the proposed DRL approach outperforms benchmarks and solves the large-scale WNCS codesign problem, providing a scalable solution for industrial automation.
Authors: Yunho Kim, Jaehyun Park, Heejun Kim, Sejin Kim, Byung-Jun Lee, Sundong Kim
Abstract: Effective long-term strategies enable AI systems to navigate complex environments by making sequential decisions over extended horizons. Similarly, reinforcement learning (RL) agents optimize decisions across sequences to maximize rewards, even without immediate feedback. To verify that Latent Diffusion-Constrained Q-learning (LDCQ), a prominent diffusion-based offline RL method, demonstrates strong reasoning abilities in multi-step decision-making, we aimed to evaluate its performance on the Abstraction and Reasoning Corpus (ARC). However, applying offline RL methodologies to enhance strategic reasoning in AI for solving tasks in ARC is challenging due to the lack of sufficient experience data in the ARC training set. To address this limitation, we introduce an augmented offline RL dataset for ARC, called Synthesized Offline Learning Data for Abstraction and Reasoning (SOLAR), along with the SOLAR-Generator, which generates diverse trajectory data based on predefined rules. SOLAR enables the application of offline RL methods by offering sufficient experience data. We synthesized SOLAR for a simple task and used it to train an agent with the LDCQ method. Our experiments demonstrate the effectiveness of the offline RL approach on a simple ARC task, showing the agent's ability to make multi-step sequential decisions and correctly identify answer states. These results highlight the potential of the offline RL approach to enhance AI's strategic reasoning capabilities.
Authors: Han Liu, Xianfeng Tang, Tianlang Chen, Jiapeng Liu, Indu Indu, Henry Peng Zou, Peng Dai, Roberto Fernandez Galan, Michael D Porter, Dongmei Jia, Ning Zhang, Lian Xiong
Abstract: The fashion industry is one of the leading domains in the global e-commerce sector, prompting major online retailers to employ recommendation systems for product suggestions and customer convenience. While recommendation systems have been widely studied, most are designed for general e-commerce problems and struggle with the unique challenges of the fashion domain. To address these issues, we propose a sequential fashion recommendation framework that leverages a pre-trained large language model (LLM) enhanced with recommendation-specific prompts. Our framework employs parameter-efficient fine-tuning with extensive fashion data and introduces a novel mix-up-based retrieval technique for translating text into relevant product suggestions. Extensive experiments show our proposed framework significantly enhances fashion recommendation performance.
Authors: Syed Abdul Gaffar Shakhadri, Kruthika KR, Rakshit Aralimatti
Abstract: We introduce Shakti, a 2.5 billion parameter language model specifically optimized for resource-constrained environments such as edge devices, including smartphones, wearables, and IoT systems. Shakti combines high-performance NLP with optimized efficiency and precision, making it ideal for real-time AI applications where computational resources and memory are limited. With support for vernacular languages and domain-specific tasks, Shakti excels in industries such as healthcare, finance, and customer service. Benchmark evaluations demonstrate that Shakti performs competitively against larger models while maintaining low latency and on-device efficiency, positioning it as a leading solution for edge AI.
Authors: Lap Chi Lau, Robert Wang, Hong Zhou
Abstract: We present a unified deterministic approach for experimental design problems using the method of interlacing polynomials. Our framework recovers the best-known approximation guarantees for the well-studied D/A/E-design problems with simple analysis. Furthermore, we obtain improved non-trivial approximation guarantee for E-design in the challenging small budget regime. Additionally, our approach provides an optimal approximation guarantee for a generalized ratio objective that generalizes both D-design and A-design.
Authors: Vivin Vinod, Peter Zaspel
Abstract: The development of machine learning (ML) methods has made quantum chemistry (QC) calculations more accessible by reducing the compute cost incurred in conventional QC methods. This has since been translated into the overhead cost of generating training data. Increased work in reducing the cost of generating training data resulted in the development of $\Delta$-ML and multifidelity machine learning methods which use data at more than one QC level of accuracy, or fidelity. This work compares the data costs associated with $\Delta$-ML, multifidelity machine learning (MFML), and optimized MFML (o-MFML) in contrast with a newly introduced Multifidelity$\Delta$-Machine Learning (MF$\Delta$ML) method for the prediction of ground state energies over the multifidelity benchmark dataset QeMFi. This assessment is made on the basis of training data generation cost associated with each model and is compared with the single fidelity kernel ridge regression (KRR) case. The results indicate that the use of multifidelity methods surpasses the standard $\Delta$-ML approaches in cases of a large number of predictions. For cases, where $\Delta$-ML method might be favored, such as small test set regimes, the MF$\Delta$-ML method is shown to be more efficient than conventional $\Delta$-ML.
Authors: Vivin Vinod, Peter Zaspel
Abstract: Recent progress in machine learning (ML) has made high-accuracy quantum chemistry (QC) calculations more accessible. Of particular interest are multifidelity machine learning (MFML) methods where training data from differing accuracies or fidelities are used. These methods usually employ a fixed scaling factor, $\gamma$, to relate the number of training samples across different fidelities, which reflects the cost and assumed sparsity of the data. This study investigates the impact of modifying $\gamma$ on model efficiency and accuracy for the prediction of vertical excitation energies using the QeMFi benchmark dataset. Further, this work introduces QC compute time informed scaling factors, denoted as $\theta$, that vary based on QC compute times at different fidelities. A novel error metric, error contours of MFML, is proposed to provide a comprehensive view of model error contributions from each fidelity. The results indicate that high model accuracy can be achieved with just 2 training samples at the target fidelity when a larger number of samples from lower fidelities are used. This is further illustrated through a novel concept, the $\Gamma$-curve, which compares model error against the time-cost of generating training samples, demonstrating that multifidelity models can achieve high accuracy while minimizing training data costs.
Authors: Hanti Lin
Abstract: This article reviews and develops an epistemological tradition in philosophy of science, called convergentism, which holds that inference methods should be assessed in terms of their abilities to converge to the truth. This tradition is compared with three competing ones: (1) explanationism, which holds that theory choice should be guided by a theory's overall balance of explanatory virtues, such as simplicity and fit with data; (2) instrumentalism, according to which scientific inference should be driven by the goal of obtaining useful models, rather than true theories; (3) Bayesianism, which features a shift of focus from all-or-nothing beliefs to degrees of belief.
Authors: Jingtao Guo, Wenhao Zhuang, Yuyi Mao, Ivan Wang-Hei Ho
Abstract: Passenger counting is crucial for public transport vehicle scheduling and traffic capacity evaluation. However, most existing methods are either costly or with low counting accuracy, leading to the recent use of Wi-Fi signals for this purpose. In this paper, we develop an efficient edge computing-based passenger counting system consists of multiple Wi-Fi receivers and an edge server. It leverages channel state information (CSI) and received signal strength indicator (RSSI) to facilitate the collaboration among multiple receivers. Specifically, we design a novel CSI feature fusion module called Adaptive RSSI-weighted CSI Feature Concatenation, which integrates locally extracted CSI and RSSI features from multiple receivers for information fusion at the edge server. Performance of our proposed system is evaluated using a real-world dataset collected from a double-decker bus in Hong Kong, with up to 20 passengers. The experimental results reveal that our system achieves an average accuracy and F1-score of over 94%, surpassing other cooperative sensing baselines by at least 2.27% in accuracy and 2.34% in F1-score.
Authors: Kenji Ueda, Masaki Nishiura
Abstract: We propose a novel tomographic method, nonlinear Gaussian process tomography (nonlinear GPT) that employs the Laplace approximation to ensure the non-negative physical quantity, such as the emissivity of plasma optical diagnostics. This new method implements a logarithmic Gaussian process (log-GP) to model plasma distribution more naturally, thereby expanding the limitations of standard GPT, which are restricted to linear problems and may yield non-physical negative values. The effectiveness of the proposed log-GP tomography is demonstrated through a case study using the Ring Trap 1 (RT-1) device, where log-GPT outperforms existing methods, standard GPT, and the Minimum Fisher Information (MFI) methods in terms of reconstruction accuracy. The result highlights the effectiveness of nonlinear GPT for imposing physical constraints in applications to an inverse problem.
Authors: Animesh Singh Basnet, Mohamed Chahine Ghanem, Dipo Dunsin, Wiktor Sowinski-Mydlarz
Abstract: This paper investigates the application of Deep Reinforcement Learning (DRL) for attributing malware to specific Advanced Persistent Threat (APT) groups through detailed behavioural analysis. By analysing over 3500 malware samples from 12 distinct APT groups, the study utilises sophisticated tools like Cuckoo Sandbox to extract behavioural data, providing a deep insight into the operational patterns of malware. The research demonstrates that the DRL model significantly outperforms traditional machine learning approaches such as SGD, SVC, KNN, MLP, and Decision Tree Classifiers, achieving an impressive test accuracy of 89.27 %. It highlights the model capability to adeptly manage complex, variable, and elusive malware attributes. Furthermore, the paper discusses the considerable computational resources and extensive data dependencies required for deploying these advanced AI models in cybersecurity frameworks. Future research is directed towards enhancing the efficiency of DRL models, expanding the diversity of the datasets, addressing ethical concerns, and leveraging Large Language Models (LLMs) to refine reward mechanisms and optimise the DRL framework. By showcasing the transformative potential of DRL in malware attribution, this research advocates for a responsible and balanced approach to AI integration, with the goal of advancing cybersecurity through more adaptable, accurate, and robust systems.
Authors: Yi Sun, Yuri M. Brovman
Abstract: There are unique challenges to developing item recommender systems for e-commerce platforms like eBay due to sparse data and diverse user interests. While rich user-item interactions are important, eBay's data sparsity exceeds other e-commerce sites by an order of magnitude. To address this challenge, we propose CoActionGraphRec (CAGR), a text based two-tower deep learning model (Item Tower and User Tower) utilizing co-action graph layers. In order to enhance user and item representations, a graph-based solution tailored to eBay's environment is utilized. For the Item Tower, we represent each item using its co-action items to capture collaborative signals in a co-action graph that is fully leveraged by the graph neural network component. For the User Tower, we build a fully connected graph of each user's behavior sequence, with edges encoding pairwise relationships. Furthermore, an explicit interaction module learns representations capturing behavior interactions. Extensive offline and online A/B test experiments demonstrate the effectiveness of our proposed approach and results show improved performance over state-of-the-art methods on key metrics.
Authors: Guixian Zhang, Guan Yuan, Debo Cheng, Lin Liu, Jiuyong Li, Shichao Zhang
Abstract: With the widespread use of Graph Neural Networks (GNNs) for representation learning from network data, the fairness of GNN models has raised great attention lately. Fair GNNs aim to ensure that node representations can be accurately classified, but not easily associated with a specific group. Existing advanced approaches essentially enhance the generalisation of node representation in combination with data augmentation strategy, and do not directly impose constraints on the fairness of GNNs. In this work, we identify that a fundamental reason for the unfairness of GNNs in social network learning is the phenomenon of social homophily, i.e., users in the same group are more inclined to congregate. The message-passing mechanism of GNNs can cause users in the same group to have similar representations due to social homophily, leading model predictions to establish spurious correlations with sensitive attributes. Inspired by this reason, we propose a method called Equity-Aware GNN (EAGNN) towards fair graph representation learning. Specifically, to ensure that model predictions are independent of sensitive attributes while maintaining prediction performance, we introduce constraints for fair representation learning based on three principles: sufficiency, independence, and separation. We theoretically demonstrate that our EAGNN method can effectively achieve group fairness. Extensive experiments on three datasets with varying levels of social homophily illustrate that our EAGNN method achieves the state-of-the-art performance across two fairness metrics and offers competitive effectiveness.
Authors: Weixi Xiang, Xueting Han, Xiujuan Chai, Jing Bai
Abstract: Modeling biological sequences such as DNA, RNA, and proteins is crucial for understanding complex processes like gene regulation and protein synthesis. However, most current models either focus on a single type or treat multiple types of data separately, limiting their ability to capture cross-modal relationships. We propose that by learning the relationships between these modalities, the model can enhance its understanding of each type. To address this, we introduce BSM, a small but powerful mixed-modal biological sequence foundation model, trained on three types of data: RefSeq, Gene Related Sequences, and interleaved biological sequences from the web. These datasets capture the genetic flow, gene-protein relationships, and the natural co-occurrence of diverse biological data, respectively. By training on mixed-modal data, BSM significantly enhances learning efficiency and cross-modal representation, outperforming models trained solely on unimodal data. With only 110M parameters, BSM achieves performance comparable to much larger models across both single-modal and mixed-modal tasks, and uniquely demonstrates in-context learning capability for mixed-modal tasks, which is absent in existing models. Further scaling to 270M parameters demonstrates even greater performance gains, highlighting the potential of BSM as a significant advancement in multimodal biological sequence modeling.
Authors: Lan V. Truong
Abstract: In this paper, we introduce various covering number bounds for linear function classes, each subject to different constraints on input and matrix norms. These bounds are contingent on the rank of each class of matrices. We then apply these bounds to derive generalization errors for single layer transformers. Our results improve upon several existing generalization bounds in the literature and are independent of input sequence length, highlighting the advantages of employing low-rank matrices in transformer design. More specifically, our achieved generalisation error bound decays as $O(1/\sqrt{n})$ where $n$ is the sample length, which improves existing results in research literature of the order $O((\log n)/(\sqrt{n}))$. It also decays as $O(\log r_w)$ where $r_w$ is the rank of the combination of query and and key matrices.
Authors: Renhang Liu, Abhinaba Roy, Dorien Herremans
Abstract: In this work, we present a novel method for music emotion recognition that leverages Large Language Model (LLM) embeddings for label alignment across multiple datasets and zero-shot prediction on novel categories. First, we compute LLM embeddings for emotion labels and apply non-parametric clustering to group similar labels, across multiple datasets containing disjoint labels. We use these cluster centers to map music features (MERT) to the LLM embedding space. To further enhance the model, we introduce an alignment regularization that enables dissociation of MERT embeddings from different clusters. This further enhances the model's ability to better adaptation to unseen datasets. We demonstrate the effectiveness of our approach by performing zero-shot inference on a new dataset, showcasing its ability to generalize to unseen labels without additional training.
Authors: Jeff Guo, Philippe Schwaller
Abstract: Constrained synthesizability is an unaddressed challenge in generative molecular design. In particular, designing molecules satisfying multi-parameter optimization objectives, while simultaneously being synthesizable and enforcing the presence of specific commercial building blocks in the synthesis. This is practically important for molecule re-purposing, sustainability, and efficiency. In this work, we propose a novel reward function called TANimoto Group Overlap (TANGO), which uses chemistry principles to transform a sparse reward function into a dense and learnable reward function -- crucial for reinforcement learning. TANGO can augment general-purpose molecular generative models to directly optimize for constrained synthesizability while simultaneously optimizing for other properties relevant to drug discovery using reinforcement learning. Our framework is general and addresses starting-material, intermediate, and divergent synthesis constraints. Contrary to most existing works in the field, we show that incentivizing a general-purpose (without any inductive biases) model is a productive approach to navigating challenging optimization scenarios. We demonstrate this by showing that the trained models explicitly learn a desirable distribution. Our framework is the first generative approach to tackle constrained synthesizability.
Authors: Giovanni Ziarelli, Stefano Pagani, Nicola Parolini, Francesco Regazzoni, Marco Verani
Abstract: In this work, we aim to formalize a novel scientific machine learning framework to reconstruct the hidden dynamics of the transmission rate, whose inaccurate extrapolation can significantly impair the quality of the epidemic forecasts, by incorporating the influence of exogenous variables (such as environmental conditions and strain-specific characteristics). We propose an hybrid model that blends a data-driven layer with a physics-based one. The data-driven layer is based on a neural ordinary differential equation that learns the dynamics of the transmission rate, conditioned on the meteorological data and wave-specific latent parameters. The physics-based layer, instead, consists of a standard SEIR compartmental model, wherein the transmission rate represents an input. The learning strategy follows an end-to-end approach: the loss function quantifies the mismatch between the actual numbers of infections and its numerical prediction obtained from the SEIR model incorporating as an input the transmission rate predicted by the neural ordinary differential equation. We validate this original approach using both a synthetic test case and a realistic test case based on meteorological data (temperature and humidity) and influenza data from Italy between 2010 and 2020. In both scenarios, we achieve low generalization error on the test set and observe strong alignment between the reconstructed model and established findings on the influence of meteorological factors on epidemic spread. Finally, we implement a data assimilation strategy to adapt the neural equation to the specific characteristics of an epidemic wave under investigation, and we conduct sensitivity tests on the network hyperparameters.
Authors: Shuntaro Takahashi, Takuya Wakisaka, Hiroyuki Tokunaga
Abstract: The edge-device environment imposes severe resource limitations, encompassing computation costs, hardware resource usage, and energy consumption for deploying deep neural network models. Ultra-low-bit quantization and hardware accelerators have been explored as promising approaches to address these challenges. Ultra-low-bit quantization significantly reduces the model size and the computational cost. Despite progress so far, many competitive ultra-low-bit models still partially rely on float or non-ultra-low-bit quantized computation such as the input and output layer. We introduce Efficiera Residual Networks (ERNs), a model optimized for low-resource edge devices. ERNs achieve full ultra-low-bit quantization, with all weights, including the initial and output layers, being binary, and activations set at 2 bits. We introduce the shared constant scaling factor technique to enable integer-valued computation in residual connections, allowing our model to operate without float values until the final convolution layer. Demonstrating competitiveness, ERNs achieve an ImageNet top-1 accuracy of 72.5pt with a ResNet50-compatible architecture and 63.6pt with a model size less than 1MB. Moreover, ERNs exhibit impressive inference times, reaching 300FPS with the smallest model and 60FPS with the largest model on a cost-efficient FPGA device.
Authors: Chunlin Tian, Li Li, Kahou Tam, Yebo Wu, Chengzhong Xu
Abstract: Federated Learning (FL) enables multiple devices to collaboratively train a shared model while preserving data privacy. Ever-increasing model complexity coupled with limited memory resources on the participating devices severely bottlenecks the deployment of FL in real-world scenarios. Thus, a framework that can effectively break the memory wall while jointly taking into account the hardware and statistical heterogeneity in FL is urgently required. In this paper, we propose SmartSplit, a framework that effectively reduces the memory footprint on the device side while guaranteeing the training progress and model accuracy for heterogeneous FL through model splitting.Towards this end, SmartSplit employs a hierarchical structure to adaptively guide the overall training process. In each training round, the central manager, hosted on the server, dynamically selects the participating devices and sets the cutting layer by jointly considering the memory budget, training capacity, and data distribution of each device. The MEC manager, deployed within the edge server, proceeds to split the local model and perform training of the server-side portion. Meanwhile, it fine-tunes the splitting points based on the time-evolving statistical importance. The on-device manager, embedded inside each mobile device, continuously monitors the local training status while employing cost-aware checkpointing to match the runtime dynamic memory budget. Extensive experiments on representative datasets are conducted on both commercial off-the-shelf mobile device testbeds. The experimental results show that SmartSplit excels in FL training on highly memory-constrained mobile SoCs, offering up to a 94% peak latency reduction and 100-fold memory savings. It enhances accuracy performance by 1.49%-57.18% and adaptively adjusts to dynamic memory budgets through cost-aware recomputation.
Authors: Yake Wei, Di Hu, Henghui Du, Ji-Rong Wen
Abstract: Multimodal learning is expected to boost model performance by integrating information from different modalities. However, its potential is not fully exploited because the widely-used joint training strategy, which has a uniform objective for all modalities, leads to imbalanced and under-optimized uni-modal representations. Specifically, we point out that there often exists modality with more discriminative information, e.g., vision of playing football and sound of blowing wind. They could dominate the joint training process, resulting in other modalities being significantly under-optimized. To alleviate this problem, we first analyze the under-optimized phenomenon from both the feed-forward and the back-propagation stages during optimization. Then, On-the-fly Prediction Modulation (OPM) and On-the-fly Gradient Modulation (OGM) strategies are proposed to modulate the optimization of each modality, by monitoring the discriminative discrepancy between modalities during training. Concretely, OPM weakens the influence of the dominant modality by dropping its feature with dynamical probability in the feed-forward stage, while OGM mitigates its gradient in the back-propagation stage. In experiments, our methods demonstrate considerable improvement across a variety of multimodal tasks. These simple yet effective strategies not only enhance performance in vanilla and task-oriented multimodal models, but also in more complex multimodal tasks, showcasing their effectiveness and flexibility. The source code is available at \url{https://github.com/GeWu-Lab/BML_TPAMI2024}.
Authors: Manuel Barusco, Francesco Borsatti, Davide Dalle Pezze, Francesco Paissan, Elisabetta Farella, Gian Antonio Susto
Abstract: Visual Anomaly Detection (VAD) has gained significant research attention for its ability to identify anomalous images and pinpoint the specific areas responsible for the anomaly. A key advantage of VAD is its unsupervised nature, which eliminates the need for costly and time-consuming labeled data collection. However, despite its potential for real-world applications, the literature has given limited focus to resource-efficient VAD, particularly for deployment on edge devices. This work addresses this gap by leveraging lightweight neural networks to reduce memory and computation requirements, enabling VAD deployment on resource-constrained edge devices. We benchmark the major VAD algorithms within this framework and demonstrate the feasibility of edge-based VAD using the well-known MVTec dataset. Furthermore, we introduce a novel algorithm, Partially Shared Teacher-student (PaSTe), designed to address the high resource demands of the existing Student Teacher Feature Pyramid Matching (STFPM) approach. Our results show that PaSTe decreases the inference time by 25%, while reducing the training time by 33% and peak RAM usage during training by 76%. These improvements make the VAD process significantly more efficient, laying a solid foundation for real-world deployment on edge devices.
Authors: Arturo Salmi, Szabolcs Cs\'efalvay, James Imber
Abstract: Despite recent advances in hardware acceleration of ray tracing, real-time ray budgets remain stubbornly limited at a handful of samples per pixel (spp) on commodity hardware, placing the onus on denoising algorithms to achieve high visual quality for path traced global illumination. Neural network-based solutions give excellent result quality at the cost of increased execution time relative to hand-engineered methods, making them less suitable for deployment on resource-constrained systems. We therefore propose incorporating a neural network into a computationally-efficient local linear model-based denoiser, and demonstrate faithful single-frame reconstruction of global illumination for Lambertian scenes at very low sample counts (1spp) and for low computational cost. Other contributions include improving the quality and performance of local linear model-based denoising through a simplified mathematical treatment, and demonstration of the surprising usefulness of ambient occlusion as a guide channel. We also show how our technique is straightforwardly extensible to joint denoising and upsampling of path traced renders with reference to low-cost, rasterized guide channels.
Authors: Chiyi Huang, Longwei Sun, Dong Liang, Haifeng Liang, Hongwu Zeng, Yanjie Zhu
Abstract: Cardiac T1 mapping can evaluate various clinical symptoms of myocardial tissue. However, there is currently a lack of effective, robust, and efficient methods for motion correction in cardiac T1 mapping. In this paper, we propose a deep learning-based and topology-preserving image registration framework for motion correction in cardiac T1 mapping. Notably, our proposed implicit consistency constraint dubbed BLOC, to some extent preserves the image topology in registration by bidirectional consistency constraint and local anti-folding constraint. To address the contrast variation issue, we introduce a weighted image similarity metric for multimodal registration of cardiac T1-weighted images. Besides, a semi-supervised myocardium segmentation network and a dual-domain attention module are integrated into the framework to further improve the performance of the registration. Numerous comparative experiments, as well as ablation studies, demonstrated the effectiveness and high robustness of our method. The results also indicate that the proposed weighted image similarity metric, specifically crafted for our network, contributes a lot to the enhancement of the motion correction efficacy, while the bidirectional consistency constraint combined with the local anti-folding constraint ensures a more desirable topology-preserving registration mapping.
Authors: Federico Pizarro Bejarano, Lukas Brunke, Angela P. Schoellig
Abstract: Reinforcement learning (RL) controllers are flexible and performant but rarely guarantee safety. Safety filters impart hard safety guarantees to RL controllers while maintaining flexibility. However, safety filters can cause undesired behaviours due to the separation between the controller and the safety filter, often degrading performance and robustness. In this paper, we propose several modifications to incorporating the safety filter in training RL controllers rather than solely applying it during evaluation. The modifications allow the RL controller to learn to account for the safety filter, improving performance. Additionally, our modifications significantly improve sample efficiency and eliminate training-time constraint violations. We verified the proposed modifications in simulated and real experiments with a Crazyflie 2.0 drone. In experiments, we show that the proposed training approaches require significantly fewer environment interactions and improve performance by up to 20% compared to standard RL training.
Authors: Zhengyan Shi, Sander Land, Acyr Locatelli, Matthieu Geist, Max Bartolo
Abstract: Direct Alignment Algorithms (DAAs), such as Direct Preference Optimisation (DPO) and Identity Preference Optimisation (IPO), have emerged as alternatives to online Reinforcement Learning from Human Feedback (RLHF) algorithms such as Proximal Policy Optimisation (PPO) for aligning language models to human preferences, without the need for explicit reward modelling. These methods generally aim to increase the likelihood of generating better (preferred) completions while discouraging worse (non-preferred) ones, while staying close to the original model's behaviour. In this work, we explore the relationship between completion likelihood and model performance in state-of-the-art DAAs, and identify a critical issue of likelihood over-optimisation. Contrary to expectations, we find that higher likelihood of better completions and larger margins between better and worse completion likelihoods do not necessarily lead to better performance, and may even degrade it. Our analysis reveals that while higher likelihood correlates with better memorisation of factual knowledge patterns, a slightly lower completion likelihood tends to improve output diversity, thus leading to better generalisation to unseen scenarios. Moreover, we identify two key indicators that signal when over-optimised output diversity begins to harm performance: Decreasing Entropy over Top-k Tokens and Diminishing Top-k Probability Mass. Our experimental results validate that these indicators are reliable signs of declining performance under different regularisations, helping prevent over-optimisation and improve alignment with human preferences.
Authors: Mar\'ia Victoria Carro, Francisca Gauna Selasco, Denise Alejandra Mester, Mario Alejandro Leiva
Abstract: Illusions of causality occur when people develop the belief that there is a causal connection between two variables with no supporting evidence. This cognitive bias has been proposed to underlie many societal problems including social prejudice, stereotype formation, misinformation and superstitious thinking. In this research we investigate whether large language models develop the illusion of causality in real-world settings. We evaluated and compared news headlines generated by GPT-4o-Mini, Claude-3.5-Sonnet, and Gemini-1.5-Pro to determine whether the models incorrectly framed correlations as causal relationships. In order to also measure sycophantic behavior, which occurs when a model aligns with a user's beliefs in order to look favorable even if it is not objectively correct, we additionally incorporated the bias into the prompts, observing if this manipulation increases the likelihood of the models exhibiting the illusion of causality. We found that Claude-3.5-Sonnet is the model that presents the lowest degree of causal illusion aligned with experiments on Correlation-to-Causation Exaggeration in human-written press releases. On the other hand, our findings suggest that while mimicry sycophancy increases the likelihood of causal illusions in these models, especially in GPT-4o-Mini, Claude-3.5-Sonnet remains the most robust against this cognitive bias.
Authors: Abdelhakim Benechehab, Youssef Attia El Hili, Ambroise Odonnat, Oussama Zekri, Albert Thomas, Giuseppe Paolo, Maurizio Filippone, Ievgen Redko, Bal\'azs K\'egl
Abstract: The emerging zero-shot capabilities of Large Language Models (LLMs) have led to their applications in areas extending well beyond natural language processing tasks. In reinforcement learning, while LLMs have been extensively used in text-based environments, their integration with continuous state spaces remains understudied. In this paper, we investigate how pre-trained LLMs can be leveraged to predict in context the dynamics of continuous Markov decision processes. We identify handling multivariate data and incorporating the control signal as key challenges that limit the potential of LLMs' deployment in this setup and propose Disentangled In-Context Learning (DICL) to address them. We present proof-of-concept applications in two reinforcement learning settings: model-based policy evaluation and data-augmented off-policy reinforcement learning, supported by theoretical analysis of the proposed methods. Our experiments further demonstrate that our approach produces well-calibrated uncertainty estimates. We release the code at https://github.com/abenechehab/dicl.
Authors: Yuhang Liang, Xinyi Li, Jie Ren, Ang Li, Bo Fang, Jieyang Chen
Abstract: Large Language Models (LLMs) have demonstrated remarkable performance in various natural language processing tasks. However, the training of these models is computationally intensive and susceptible to faults, particularly in the attention mechanism, which is a critical component of transformer-based LLMs. In this paper, we investigate the impact of faults on LLM training, focusing on INF, NaN, and near-INF values in the computation results with systematic fault injection experiments. We observe the propagation patterns of these errors, which can trigger non-trainable states in the model and disrupt training, forcing the procedure to load from checkpoints.To mitigate the impact of these faults, we propose ATTNChecker, the first Algorithm-Based Fault Tolerance (ABFT) technique tailored for the attention mechanism in LLMs. ATTNChecker is designed based on fault propagation patterns of LLM and incorporates performance optimization to adapt to both system reliability and model vulnerability while providing lightweight protection for fast LLM training. Evaluations on four LLMs show that ATTNChecker on average incurs on average 7% overhead on training while detecting and correcting all extreme errors. Compared with the state-of-the-art checkpoint/restore approach, ATTNChecker reduces recovery overhead by up to 49x.
Authors: Seonghyeon Ye, Joel Jang, Byeongguk Jeon, Sejune Joo, Jianwei Yang, Baolin Peng, Ajay Mandlekar, Reuben Tan, Yu-Wei Chao, Bill Yuchen Lin, Lars Liden, Kimin Lee, Jianfeng Gao, Luke Zettlemoyer, Dieter Fox, Minjoon Seo
Abstract: We introduce Latent Action Pretraining for general Action models (LAPA), an unsupervised method for pretraining Vision-Language-Action (VLA) models without ground-truth robot action labels. Existing Vision-Language-Action models require action labels typically collected by human teleoperators during pretraining, which significantly limits possible data sources and scale. In this work, we propose a method to learn from internet-scale videos that do not have robot action labels. We first train an action quantization model leveraging VQ-VAE-based objective to learn discrete latent actions between image frames, then pretrain a latent VLA model to predict these latent actions from observations and task descriptions, and finally finetune the VLA on small-scale robot manipulation data to map from latent to robot actions. Experimental results demonstrate that our method significantly outperforms existing techniques that train robot manipulation policies from large-scale videos. Furthermore, it outperforms the state-of-the-art VLA model trained with robotic action labels on real-world manipulation tasks that require language conditioning, generalization to unseen objects, and semantic generalization to unseen instructions. Training only on human manipulation videos also shows positive transfer, opening up the potential for leveraging web-scale data for robotics foundation model.
Authors: Lev Sorokin, Damir Safin, Shiva Nejati
Abstract: Search-based software testing (SBST) is a widely adopted technique for testing complex systems with large input spaces, such as Deep Learning-enabled (DL-enabled) systems. Many SBST techniques focus on Pareto-based optimization, where multiple objectives are optimized in parallel to reveal failures. However, it is important to ensure that identified failures are spread throughout the entire failure-inducing area of a search domain and not clustered in a sub-region. This ensures that identified failures are semantically diverse and reveal a wide range of underlying causes. In this paper, we present a theoretical argument explaining why testing based on Pareto optimization is inadequate for covering failure-inducing areas within a search domain. We support our argument with empirical results obtained by applying two widely used types of Pareto-based optimization techniques, namely NSGA-II (an evolutionary algorithm) and MOPSO (a swarm-based algorithm), to two DL-enabled systems: an industrial Automated Valet Parking (AVP) system and a system for classifying handwritten digits. We measure the coverage of failure-revealing test inputs in the input space using a metric that we refer to as the Coverage Inverted Distance quality indicator. Our results show that NSGA-II and MOPSO are not more effective than a na\"ive random search baseline in covering test inputs that reveal failures. The replication package for this study is available in a GitHub repository.
Authors: Kai Yao, Penlei Gao, Lichun Li, Yuan Zhao, Xiaofeng Wang, Wei Wang, Jianke Zhu
Abstract: Parameter-Efficient Fine-Tuning (PEFT) methods have gained significant popularity for adapting pre-trained Large Language Models (LLMs) to downstream tasks, primarily due to their potential to significantly reduce memory and computational overheads. However, a common limitation in most PEFT approaches is their application of a uniform architectural design across all layers. This uniformity involves identical trainable modules and ignores the varying importance of each layer, leading to sub-optimal fine-tuning results. To overcome the above limitation and obtain better performance, we develop a novel approach, Importance-aware Sparse Tuning (IST), to fully utilize the inherent sparsity and select the most important subset of full layers with effective layer-wise importance scoring. The proposed IST is a versatile and plug-and-play technique compatible with various PEFT methods that operate on a per-layer basis. By leveraging the estimated importance scores, IST dynamically updates these selected layers in PEFT modules, leading to reduced memory demands. We further provide theoretical proof of convergence and empirical evidence of superior performance to demonstrate the advantages of IST over uniform updating strategies. Extensive experiments on a range of LLMs, PEFTs, and downstream tasks substantiate the effectiveness of our proposed method, showcasing IST's capacity to enhance existing layer-based PEFT methods. Our code is available at https://github.com/Kaiseem/IST.
Authors: Chenxi Wang, Xiang Chen, Ningyu Zhang, Bozhong Tian, Haoming Xu, Shumin Deng, Huajun Chen
Abstract: Multimodal Large Language Models (MLLMs) frequently exhibit hallucination phenomena, but the underlying reasons remain poorly understood. In this paper, we present an empirical analysis and find that, although MLLMs incorrectly generate the objects in the final output, they are actually able to recognize visual objects in the preceding layers. We speculate that this may be due to the strong knowledge priors of the language model suppressing the visual information, leading to hallucinations. Motivated by this, we propose a novel dynamic correction decoding method for MLLMs (DeCo), which adaptively selects the appropriate preceding layers and proportionally integrates knowledge into the final layer to adjust the output logits. Note that DeCo is model agnostic and can be seamlessly incorporated with various classic decoding strategies and applied to different MLLMs. We evaluate DeCo on widely-used benchmarks, demonstrating that it can reduce hallucination rates by a large margin compared to baselines, highlighting its potential to mitigate hallucinations. Code is available at https://github.com/zjunlp/DeCo.
Authors: Guibin Zhang, Yanwei Yue, Xiangguo Sun, Guancheng Wan, Miao Yu, Junfeng Fang, Kun Wang, Dawei Cheng
Abstract: Recent advancements in large language model (LLM)-based agents have demonstrated that collective intelligence can significantly surpass the capabilities of individual agents, primarily due to well-crafted inter-agent communication topologies. Despite the diverse and high-performing designs available, practitioners often face confusion when selecting the most effective pipeline for their specific task: \textit{Which topology is the best choice for my task, avoiding unnecessary communication token overhead while ensuring high-quality solution?} In response to this dilemma, we introduce G-Designer, an adaptive, efficient, and robust solution for multi-agent deployment, which dynamically designs task-aware, customized communication topologies. Specifically, G-Designer models the multi-agent system as a multi-agent network, leveraging a variational graph auto-encoder to encode both the nodes (agents) and a task-specific virtual node, and decodes a task-adaptive and high-performing communication topology. Extensive experiments on six benchmarks showcase that G-Designer is: \textbf{(1) high-performing}, achieving superior results on MMLU with accuracy at $84.50\%$ and on HumanEval with pass@1 at $89.90\%$; \textbf{(2) task-adaptive}, architecting communication protocols tailored to task difficulty, reducing token consumption by up to $95.33\%$ on HumanEval; and \textbf{(3) adversarially robust}, defending against agent adversarial attacks with merely $0.3\%$ accuracy drop.
Authors: Tsz Ting Chung, Leyang Cui, Lemao Liu, Xinting Huang, Shuming Shi, Dit-Yan Yeung
Abstract: Large Language Models (LLMs) have demonstrated impressive capabilities in a wide range of natural language processing tasks when leveraging in-context learning. To mitigate the additional computational and financial costs associated with in-context learning, several prompt compression methods have been proposed to compress the in-context learning prompts. Despite their success, these methods face challenges with transferability due to model-specific compression, or rely on external training data, such as GPT-4. In this paper, we investigate the ability of LLMs to develop a unified compression method that discretizes uninformative tokens, utilizing a self-supervised pre-training technique. By introducing a small number of parameters during the continual pre-training, the proposed Selection-p produces a probability for each input token, indicating whether to preserve or discard it. Experiments show Selection-p achieves state-of-the-art performance across numerous classification tasks, achieving compression rates of up to 10 times while experiencing only a marginal 0.8% decrease in performance. Moreover, it exhibits superior transferability to different models compared to prior work. Additionally, we further analyze how Selection-p helps maintain performance on in-context learning with long contexts.
Authors: Jinhan Li, Yifeng Zhu, Yuqi Xie, Zhenyu Jiang, Mingyo Seo, Georgios Pavlakos, Yuke Zhu
Abstract: We study the problem of teaching humanoid robots manipulation skills by imitating from single video demonstrations. We introduce OKAMI, a method that generates a manipulation plan from a single RGB-D video and derives a policy for execution. At the heart of our approach is object-aware retargeting, which enables the humanoid robot to mimic the human motions in an RGB-D video while adjusting to different object locations during deployment. OKAMI uses open-world vision models to identify task-relevant objects and retarget the body motions and hand poses separately. Our experiments show that OKAMI achieves strong generalizations across varying visual and spatial conditions, outperforming the state-of-the-art baseline on open-world imitation from observation. Furthermore, OKAMI rollout trajectories are leveraged to train closed-loop visuomotor policies, which achieve an average success rate of 79.2% without the need for labor-intensive teleoperation. More videos can be found on our website https://ut-austin-rpl.github.io/OKAMI/.
Authors: Daniel Holmberg, Emanuela Clementi, Teemu Roos
Abstract: Accurate ocean forecasting systems are vital for understanding marine dynamics, which play a crucial role in environmental management and climate adaptation strategies. Traditional numerical solvers, while effective, are computationally expensive and time-consuming. Recent advancements in machine learning have revolutionized weather forecasting, offering fast and energy-efficient alternatives. Building on these advancements, we introduce SeaCast, a neural network designed for high-resolution, medium-range ocean forecasting. SeaCast employs a graph-based framework to effectively handle the complex geometry of ocean grids and integrates external forcing data tailored to the regional ocean context. Our approach is validated through experiments at a high spatial resolution using the operational numerical model of the Mediterranean Sea provided by the Copernicus Marine Service, along with both numerical and data-driven atmospheric forcings.
Authors: Luping Liu, Chao Du, Tianyu Pang, Zehan Wang, Chongxuan Li, Dong Xu
Abstract: The rapid advancement of text-to-image (T2I) diffusion models has enabled them to generate unprecedented results from given texts. However, as text inputs become longer, existing encoding methods like CLIP face limitations, and aligning the generated images with long texts becomes challenging. To tackle these issues, we propose LongAlign, which includes a segment-level encoding method for processing long texts and a decomposed preference optimization method for effective alignment training. For segment-level encoding, long texts are divided into multiple segments and processed separately. This method overcomes the maximum input length limits of pretrained encoding models. For preference optimization, we provide decomposed CLIP-based preference models to fine-tune diffusion models. Specifically, to utilize CLIP-based preference models for T2I alignment, we delve into their scoring mechanisms and find that the preference scores can be decomposed into two components: a text-relevant part that measures T2I alignment and a text-irrelevant part that assesses other visual aspects of human preference. Additionally, we find that the text-irrelevant part contributes to a common overfitting problem during fine-tuning. To address this, we propose a reweighting strategy that assigns different weights to these two components, thereby reducing overfitting and enhancing alignment. After fine-tuning $512 \times 512$ Stable Diffusion (SD) v1.5 for about 20 hours using our method, the fine-tuned SD outperforms stronger foundation models in T2I alignment, such as PixArt-$\alpha$ and Kandinsky v2.2. The code is available at https://github.com/luping-liu/LongAlign.
Authors: Jacopo Iollo, Christophe Heinkel\'e, Pierre Alliez, Florence Forbes
Abstract: Bayesian Optimal Experimental Design (BOED) is a powerful tool to reduce the cost of running a sequence of experiments. When based on the Expected Information Gain (EIG), design optimization corresponds to the maximization of some intractable expected {\it contrast} between prior and posterior distributions. Scaling this maximization to high dimensional and complex settings has been an issue due to BOED inherent computational complexity. In this work, we introduce an {\it expected posterior} distribution with cost-effective sampling properties and provide a tractable access to the EIG contrast maximization via a new EIG gradient expression. Diffusion-based samplers are used to compute the dynamics of the expected posterior and ideas from bi-level optimization are leveraged to derive an efficient joint sampling-optimization loop, without resorting to lower bound approximations of the EIG. The resulting efficiency gain allows to extend BOED to the well-tested generative capabilities of diffusion models. By incorporating generative models into the BOED framework, we expand its scope and its use in scenarios that were previously impractical. Numerical experiments and comparison with state-of-the-art methods show the potential of the approach.
Authors: Peng Jin, Bo Zhu, Li Yuan, Shuicheng Yan
Abstract: In this work, we upgrade the multi-head attention mechanism, the core of the Transformer model, to improve efficiency while maintaining or surpassing the previous accuracy level. We show that multi-head attention can be expressed in the summation form. Drawing on the insight that not all attention heads hold equal significance, we propose Mixture-of-Head attention (MoH), a new architecture that treats attention heads as experts in the Mixture-of-Experts (MoE) mechanism. MoH has two significant advantages: First, MoH enables each token to select the appropriate attention heads, enhancing inference efficiency without compromising accuracy or increasing the number of parameters. Second, MoH replaces the standard summation in multi-head attention with a weighted summation, introducing flexibility to the attention mechanism and unlocking extra performance potential. Extensive experiments on ViT, DiT, and LLMs demonstrate that MoH outperforms multi-head attention by using only 50%-90% of the attention heads. Moreover, we demonstrate that pre-trained multi-head attention models, such as LLaMA3-8B, can be further continue-tuned into our MoH models. Notably, MoH-LLaMA3-8B achieves an average accuracy of 64.0% across 14 benchmarks, outperforming LLaMA3-8B by 2.4% by utilizing only 75% of the attention heads. We believe the proposed MoH is a promising alternative to multi-head attention and provides a strong foundation for developing advanced and efficient attention-based models.
Authors: Ke Sun, Yingnan Zhao, Wulong Liu, Bei Jiang, Linglong Kong
Abstract: The empirical success of distributional reinforcement learning (RL) highly relies on the choice of distribution divergence equipped with an appropriate distribution representation. In this paper, we propose \textit{Sinkhorn distributional RL (SinkhornDRL)}, which leverages Sinkhorn divergence, a regularized Wasserstein loss, to minimize the difference between current and target Bellman return distributions. Theoretically, we prove the contraction properties of SinkhornDRL, aligning with the interpolation nature of Sinkhorn divergence between Wasserstein distance and Maximum Mean Discrepancy (MMD). The introduced SinkhornDRL enriches the family of distributional RL algorithms, contributing to interpreting the algorithm behaviors compared with existing approaches by our investigation into their relationships. Empirically, we show that SinkhornDRL consistently outperforms or matches existing algorithms on the Atari games suite and particularly stands out in the multi-dimensional reward setting. \thanks{Code is available in \url{https://github.com/datake/SinkhornDistRL}.}.
Authors: Claudio Battiloro, Lucia Testa, Lorenzo Giusti, Stefania Sardellitti, Paolo Di Lorenzo, Sergio Barbarossa
Abstract: Graph machine learning methods excel at leveraging pairwise relations present in the data. However, graphs are unable to fully capture the multi-way interactions inherent in many complex systems. An effective way to incorporate them is to model the data on higher-order combinatorial topological spaces, such as Simplicial Complexes (SCs) or Cell Complexes. For this reason, we introduce Generalized Simplicial Attention Neural Networks (GSANs), novel neural network architectures designed to process data living on simplicial complexes using masked self-attentional layers. Hinging on topological signal processing principles, we devise a series of principled self-attention mechanisms able to process data associated with simplices of various order, such as nodes, edges, triangles, and beyond. These schemes learn how to combine data associated with neighbor simplices of consecutive order in a task-oriented fashion, leveraging on the simplicial Dirac operator and its Dirac decomposition. We also prove that GSAN satisfies two fundamental properties: permutation equivariance and simplicial-awareness. Finally, we illustrate how our approach compares favorably with other simplicial and graph models when applied to several (inductive and transductive) tasks such as trajectory prediction, missing data imputation, graph classification, and simplex prediction.
Authors: Johannes von Oswald, Maximilian Schlegel, Alexander Meulemans, Seijin Kobayashi, Eyvind Niklasson, Nicolas Zucchet, Nino Scherrer, Nolan Miller, Mark Sandler, Blaise Ag\"uera y Arcas, Max Vladymyrov, Razvan Pascanu, Jo\~ao Sacramento
Abstract: Some autoregressive models exhibit in-context learning capabilities: being able to learn as an input sequence is processed, without undergoing any parameter changes, and without being explicitly trained to do so. The origins of this phenomenon are still poorly understood. Here we analyze a series of Transformer models trained to perform synthetic sequence prediction tasks, and discover that standard next-token prediction error minimization gives rise to a subsidiary learning algorithm that adjusts the model as new inputs are revealed. We show that this process corresponds to gradient-based optimization of a principled objective function, which leads to strong generalization performance on unseen sequences. Our findings explain in-context learning as a product of autoregressive loss minimization and inform the design of new optimization-based Transformer layers.
Authors: Qiwei Di, Tao Jin, Yue Wu, Heyang Zhao, Farzad Farnoud, Quanquan Gu
Abstract: Dueling bandits is a prominent framework for decision-making involving preferential feedback, a valuable feature that fits various applications involving human interaction, such as ranking, information retrieval, and recommendation systems. While substantial efforts have been made to minimize the cumulative regret in dueling bandits, a notable gap in the current research is the absence of regret bounds that account for the inherent uncertainty in pairwise comparisons between the dueling arms. Intuitively, greater uncertainty suggests a higher level of difficulty in the problem. To bridge this gap, this paper studies the problem of contextual dueling bandits, where the binary comparison of dueling arms is generated from a generalized linear model (GLM). We propose a new SupLinUCB-type algorithm that enjoys computational efficiency and a variance-aware regret bound $\tilde O\big(d\sqrt{\sum_{t=1}^T\sigma_t^2} + d\big)$, where $\sigma_t$ is the variance of the pairwise comparison in round $t$, $d$ is the dimension of the context vectors, and $T$ is the time horizon. Our regret bound naturally aligns with the intuitive expectation in scenarios where the comparison is deterministic, the algorithm only suffers from an $\tilde O(d)$ regret. We perform empirical experiments on synthetic data to confirm the advantage of our method over previous variance-agnostic algorithms.
Authors: Siyan Zhao, John Dang, Aditya Grover
Abstract: Many applications of large language models (LLMs), ranging from chatbots to creative writing, require nuanced subjective judgments that can differ significantly across different groups. Existing alignment algorithms can be expensive to align for each group, requiring prohibitive amounts of group-specific preference data and computation for real-world use cases. We introduce Group Preference Optimization (GPO), an alignment framework that steers language models to preferences of individual groups in a few-shot manner. In GPO, we augment the base LLM with an independent transformer module trained to predict the preferences of a group for the LLM generations. For few-shot learning, we parameterize this module as an in-context autoregressive transformer and train it via meta-learning on several groups. We empirically validate the efficacy of GPO through rigorous evaluations using LLMs with varied sizes on three human opinion adaptation tasks. These tasks involve adapting to the preferences of US demographic groups, global countries, and individual users. Our results demonstrate that GPO not only aligns models more accurately but also requires fewer group-specific preferences, and less training and inference computing resources, outperforming existing strategies such as in-context steering and fine-tuning methods.
Authors: Mufei Li, Eleonora Krea\v{c}i\'c, Vamsi K. Potluru, Pan Li
Abstract: Large-scale graphs with node attributes are increasingly common in various real-world applications. Creating synthetic, attribute-rich graphs that mirror real-world examples is crucial, especially for sharing graph data for analysis and developing learning models when original data is restricted to be shared. Traditional graph generation methods are limited in their capacity to handle these complex structures. Recent advances in diffusion models have shown potential in generating graph structures without attributes and smaller molecular graphs. However, these models face challenges in generating large attributed graphs due to the complex attribute-structure correlations and the large size of these graphs. This paper introduces a novel diffusion model, GraphMaker, specifically designed for generating large attributed graphs. We explore various combinations of node attribute and graph structure generation processes, finding that an asynchronous approach more effectively captures the intricate attribute-structure correlations. We also address scalability issues through edge mini-batching generation. To demonstrate the practicality of our approach in graph data dissemination, we introduce a new evaluation pipeline. The evaluation demonstrates that synthetic graphs generated by GraphMaker can be used to develop competitive graph machine learning models for the tasks defined over the original graphs without actually accessing these graphs, while many leading graph generation methods fall short in this evaluation.
Authors: Subhojeet Pramanik, Esraa Elelimy, Marlos C. Machado, Adam White
Abstract: In this paper we investigate transformer architectures designed for partially observable online reinforcement learning. The self-attention mechanism in the transformer architecture is capable of capturing long-range dependencies and it is the main reason behind its effectiveness in processing sequential data. Nevertheless, despite their success, transformers have two significant drawbacks that still limit their applicability in online reinforcement learning: (1) in order to remember all past information, the self-attention mechanism requires access to the whole history to be provided as context. (2) The inference cost in transformers is expensive. In this paper, we introduce recurrent alternatives to the transformer self-attention mechanism that offer context-independent inference cost, leverage long-range dependencies effectively, and performs well in online reinforcement learning task. We quantify the impact of the different components of our architecture in a diagnostic environment and assess performance gains in 2D and 3D pixel-based partially-observable environments (e.g. T-Maze, Mystery Path, Craftax, and Memory Maze). Compared with a state-of-the-art architecture, GTrXL, inference in our approach is at least 40% cheaper while reducing memory use more than 50%. Our approach either performs similarly or better than GTrXL, improving more than 37% upon GTrXL performance in harder tasks.
Authors: Prajwal Koirala, Cody Fleming
Abstract: This study presents a novel approach to addressing offline reinforcement learning (RL) problems by reframing them as regression tasks that can be effectively solved using Decision Trees. Mainly, we introduce two distinct frameworks: return-conditioned and return-weighted decision tree policies (RCDTP and RWDTP), both of which achieve notable speed in agent training as well as inference, with training typically lasting less than a few minutes. Despite the simplification inherent in this reformulated approach to offline RL, our agents demonstrate performance that is at least on par with the established methods. We evaluate our methods on D4RL datasets for locomotion and manipulation, as well as other robotic tasks involving wheeled and flying robots. Additionally, we assess performance in delayed/sparse reward scenarios and highlight the explainability of these policies through action distribution and feature importance.
Authors: Alice Bizeul, Bernhard Sch\"olkopf, Carl Allen
Abstract: In self-supervised learning (SSL), representations are learned via an auxiliary task without annotated labels. A common task is to classify augmentations or different modalities of the data, which share semantic content (e.g. an object in an image) but differ in style (e.g. the object's location). Many approaches to self-supervised learning have been proposed, e.g. SimCLR, CLIP, and DINO, which have recently gained much attention for their representations achieving downstream performance comparable to supervised learning. However, a theoretical understanding of self-supervised methods eludes. Addressing this, we present a generative latent variable model for self-supervised learning and show that several families of discriminative SSL, including contrastive methods, induce a comparable distribution over representations, providing a unifying theoretical framework for these methods. The proposed model also justifies connections drawn to mutual information and the use of a ''projection head''. Learning representations by fitting the model generatively (termed SimVAE) improves performance over discriminative and other VAE-based methods on simple image benchmarks and significantly narrows the gap between generative and discriminative representation learning in more complex settings. Importantly, as our analysis predicts, SimVAE outperforms self-supervised learning where style information is required, taking an important step toward understanding self-supervised methods and achieving task-agnostic representations.
Authors: Danil Akhtiamov, David Bosch, Reza Ghane, K Nithin Varma, Babak Hassibi
Abstract: A celebrated result by Gordon allows one to compare the min-max behavior of two Gaussian processes if certain inequality conditions are met. The consequences of this result include the Gaussian min-max (GMT) and convex Gaussian min-max (CGMT) theorems which have had far-reaching implications in high-dimensional statistics, machine learning, non-smooth optimization, and signal processing. Both theorems rely on a pair of Gaussian processes, first identified by Slepian, that satisfy Gordon's comparison inequalities. In this paper, we identify such a new pair. The resulting theorems extend the classical GMT and CGMT Theorems from the case where the underlying Gaussian matrix in the primary process has iid rows to where it has independent but non-identically-distributed ones. The new CGMT is applied to the problems of multi-source Gaussian regression, as well as to binary classification of general Gaussian mixture models.
Authors: Hezhe Qiao, Qingsong Wen, Xiaoli Li, Ee-Peng Lim, Guansong Pang
Abstract: This work considers a practical semi-supervised graph anomaly detection (GAD) scenario, where part of the nodes in a graph are known to be normal, contrasting to the extensively explored unsupervised setting with a fully unlabeled graph. We reveal that having access to the normal nodes, even just a small percentage of normal nodes, helps enhance the detection performance of existing unsupervised GAD methods when they are adapted to the semi-supervised setting. However, their utilization of these normal nodes is limited. In this paper, we propose a novel Generative GAD approach (namely GGAD) for the semi-supervised scenario to better exploit the normal nodes. The key idea is to generate pseudo anomaly nodes, referred to as 'outlier nodes', for providing effective negative node samples in training a discriminative one-class classifier. The main challenge here lies in the lack of ground truth information about real anomaly nodes. To address this challenge, GGAD is designed to leverage two important priors about the anomaly nodes -- asymmetric local affinity and egocentric closeness -- to generate reliable outlier nodes that assimilate anomaly nodes in both graph structure and feature representations. Comprehensive experiments on six real-world GAD datasets are performed to establish a benchmark for semi-supervised GAD and show that GGAD substantially outperforms state-of-the-art unsupervised and semi-supervised GAD methods with varying numbers of training normal nodes. Code will be made available at https://github.com/mala-lab/GGAD.
Authors: Paul Krzakala, Junjie Yang, R\'emi Flamary, Florence d'Alch\'e-Buc, Charlotte Laclau, Matthieu Labeau
Abstract: We propose Any2graph, a generic framework for end-to-end Supervised Graph Prediction (SGP) i.e. a deep learning model that predicts an entire graph for any kind of input. The framework is built on a novel Optimal Transport loss, the Partially-Masked Fused Gromov-Wasserstein, that exhibits all necessary properties (permutation invariance, differentiability and scalability) and is designed to handle any-sized graphs. Numerical experiments showcase the versatility of the approach that outperform existing competitors on a novel challenging synthetic dataset and a variety of real-world tasks such as map construction from satellite image (Sat2Graph) or molecule prediction from fingerprint (Fingerprint2Graph).
Authors: Spencer Hutchinson, Tianyi Chen, Mahnoosh Alizadeh
Abstract: We study the problem of online convex optimization (OCO) under unknown linear constraints that are either static, or stochastically time-varying. For this problem, we introduce an algorithm that we term Optimistically Safe OCO (OSOCO) and show that it enjoys $\tilde{O}(\sqrt{T})$ regret and no constraint violation. In the case of static linear constraints, this improves on the previous best known $\tilde{O}(T^{2/3})$ regret under the same assumptions. In the case of stochastic time-varying constraints, our work supplements existing results that show $O(\sqrt{T})$ regret and $O(\sqrt{T})$ cumulative violation under more general convex constraints and a different set of assumptions. In addition to our theoretical guarantees, we also give numerical results that further validate the effectiveness of our approach.
Authors: Chaoxi Niu, Guansong Pang, Ling Chen
Abstract: Real-life graph data often expands continually, rendering the learning of graph neural networks (GNNs) on static graph data impractical. Graph continual learning (GCL) tackles this problem by continually adapting GNNs to the expanded graph of the current task while maintaining the performance over the graph of previous tasks. Memory replay-based methods, which aim to replay data of previous tasks when learning new tasks, have been explored as one principled approach to mitigate the forgetting of the knowledge learned from the previous tasks. In this paper we extend this methodology with a novel framework, called Debiased Lossless Memory replay (DeLoMe). Unlike existing methods that sample nodes/edges of previous graphs to construct the memory, DeLoMe learns small lossless synthetic node representations as the memory. The learned memory can not only preserve the graph data privacy but also capture the holistic graph information, for which the sampling-based methods are not viable. Further, prior methods suffer from bias toward the current task due to the data imbalance between the classes in the memory data and the current data. A debiased GCL loss function is devised in DeLoMe to effectively alleviate this bias. Extensive experiments on four graph datasets show the effectiveness of DeLoMe under both class- and task-incremental learning settings.
Authors: Pavamana K J, Chandramani Kishore Singh
Abstract: We study content caching with recommendations in a wireless network where the users are connected through a base station equipped with a finite-capacity cache. We assume a fixed set of contents with unknown user preferences and content popularities. The base station can cache a subset of the contents and can also recommend subsets of the contents to different users in order to encourage them to request the recommended contents. Recommendations, depending on their acceptability, can thus be used to increase cache hits. We first assume that the users' recommendation acceptabilities are known and formulate the cache hit optimization problem as a combinatorial multi-armed bandit (CMAB). We propose a UCB-based algorithm to decide which contents to cache and recommend and provide an upper bound on the regret of this algorithm. Subsequently, we consider a more general scenario where the users' recommendation acceptabilities are also unknown and propose another UCB-based algorithm that learns these as well. We numerically demonstrate the performance of our algorithms and compare these to state-of-the-art algorithms.
Authors: Shi Yin, Xinyang Pan, Fengyan Wang, Lixin He
Abstract: We propose both a theoretical and a methodological framework to address a critical challenge in applying deep learning to physical systems: the reconciliation of non-linear expressiveness with SO(3)-equivariance in predictions of SO(3)-equivariant quantities. Inspired by covariant theory in physics, we present a solution by exploring the mathematical relationships between SO(3)-invariant and SO(3)-equivariant quantities and their representations. We first construct theoretical SO(3)-invariant quantities derived from the SO(3)-equivariant regression targets, and use these invariant quantities as supervisory labels to guide the learning of high-quality SO(3)-invariant features. Given that SO(3)-invariance is preserved under non-linear operations, the encoding process for invariant features can extensively utilize non-linear mappings, thereby fully capturing the non-linear patterns inherent in physical systems. Building on this, we propose a gradient-based mechanism to induce SO(3)-equivariant encodings of various degrees from the learned SO(3)-invariant features. This mechanism can incorporate non-linear expressive capabilities into SO(3)-equivariant representations, while theoretically preserving their equivariant properties as we prove, establishing a strong foundation for regressing complex SO(3)-equivariant targets. We apply our theory and method to the electronic-structure Hamiltonian prediction tasks, experimental results on eight benchmark databases covering multiple types of systems and challenging scenarios show substantial improvements on the state-of-the-art prediction accuracy of deep learning paradigm. Our method boosts Hamiltonian prediction accuracy by up to 40% and enhances downstream physical quantities, such as occupied orbital energy, by a maximum of 76%.
Authors: Christian Intern\`o, Elena Raponi, Niki van Stein, Thomas B\"ack, Markus Olhofer, Yaochu Jin, Barbara Hammer
Abstract: The rapid proliferation of smart devices coupled with the advent of 6G networks has profoundly reshaped the domain of collaborative machine learning. Alongside growing privacy-security concerns in sensitive fields, these developments have positioned federated learning (FL) as a pivotal technology for decentralized model training. Despite its vast potential, specially in the age of complex foundation models, FL encounters challenges such as elevated communication costs, computational constraints, and the complexities of non-IID data distributions. We introduce AutoFLIP, an innovative approach that utilizes a federated loss exploration phase to drive adaptive hybrid pruning, operating in a structured and unstructured way. This innovative mechanism automatically identifies and prunes model substructure by distilling knowledge on model gradients behavior across different non-IID client losses topology, thereby optimizing computational efficiency and enhancing model performance on resource constrained scenarios. Extensive experiments on various datasets and FL tasks reveal that AutoFLIP not only efficiently accelerates global convergence, but also achieves superior accuracy and robustness compared to traditional methods. On average, AutoFLIP reduces computational overhead by 48.8% and communication costs by 35.5%, while improving global accuracy. By significantly reducing these overheads, AutoFLIP offer the way for efficient FL deployment in real-world applications for a scalable and broad applicability.
Authors: Yuwei Cheng, Fan Yao, Xuefeng Liu, Haifeng Xu
Abstract: This paper studies Learning from Imperfect Human Feedback (LIHF), addressing the potential irrationality or imperfect perception when learning from comparative human feedback. Building on evidences that human's imperfection decays over time (i.e., humans learn to improve), we cast this problem as a concave-utility continuous-action dueling bandit but under a restricted form of corruption: i.e., the corruption scale is decaying over time as $t^{\rho-1}$ for some "imperfection rate" $\rho \in [0, 1]$. With $T$ as the total number of iterations, we establish a regret lower bound of $ \Omega(\max\{\sqrt{T}, T^{\rho}\}) $ for LIHF, even when $\rho$ is known. For the same setting, we develop the Robustified Stochastic Mirror Descent for Imperfect Dueling (RoSMID) algorithm, which achieves nearly optimal regret $\tilde{\mathcal{O}}(\max\{\sqrt{T}, T^{\rho}\})$. Core to our analysis is a novel framework for analyzing gradient-based algorithms for dueling bandit under corruption, and we demonstrate its general applicability by showing how this framework can be easily applied to obtain corruption-robust guarantees for other popular gradient-based dueling bandit algorithms. Our theoretical results are validated by extensive experiments.
Authors: Spencer Young, Cole Edgren, Riley Sinema, Andrew Hall, Nathan Dong, Porter Jenkins
Abstract: While significant progress has been made in specifying neural networks capable of representing uncertainty, deep networks still often suffer from overconfidence and misaligned predictive distributions. Existing approaches for addressing this misalignment are primarily developed under the framework of calibration, with common metrics such as Expected Calibration Error (ECE). However, calibration can only provide a strictly marginal assessment of probabilistic alignment. Consequently, calibration metrics such as ECE are distribution-wise measures and cannot diagnose the point-wise reliability of individual inputs, which is important for real-world decision-making. We propose a stronger condition, which we term conditional congruence, for assessing probabilistic fit. We also introduce a metric, Conditional Congruence Error (CCE), that uses conditional kernel mean embeddings to estimate the distance, at any point, between the learned predictive distribution and the empirical, conditional distribution in a dataset. We show that using CCE to measure congruence 1) accurately quantifies misalignment between distributions when the data generating process is known, 2) effectively scales to real-world, high dimensional image regression tasks, and 3) can be used to gauge model reliability on unseen instances.
Authors: Kihyun Kim, Jiawei Zhang, Asuman Ozdaglar, Pablo A. Parrilo
Abstract: Inverse Reinforcement Learning (IRL) and Reinforcement Learning from Human Feedback (RLHF) are pivotal methodologies in reward learning, which involve inferring and shaping the underlying reward function of sequential decision-making problems based on observed human demonstrations and feedback. Most prior work in reward learning has relied on prior knowledge or assumptions about decision or preference models, potentially leading to robustness issues. In response, this paper introduces a novel linear programming (LP) framework tailored for offline reward learning. Utilizing pre-collected trajectories without online exploration, this framework estimates a feasible reward set from the primal-dual optimality conditions of a suitably designed LP, and offers an optimality guarantee with provable sample efficiency. Our LP framework also enables aligning the reward functions with human feedback, such as pairwise trajectory comparison data, while maintaining computational tractability and sample efficiency. We demonstrate that our framework potentially achieves better performance compared to the conventional maximum likelihood estimation (MLE) approach through analytical examples and numerical experiments.
Authors: Ce Ge, Zhijian Ma, Daoyuan Chen, Yaliang Li, Bolin Ding
Abstract: Large language models have demonstrated remarkable capabilities across various tasks, primarily attributed to the utilization of diversely sourced data. However, the impact of pretraining data composition on model performance remains poorly understood. This paper introduces $\textbf{BiMix}$, a novel bivariate data mixing law that models the joint scaling behavior of domain proportions and data volume in LLM pretraining. $\textbf{BiMix}$ provides a systematic framework for understanding and optimizing data mixtures across diverse domains. Through extensive experiments on two large-scale datasets, we demonstrate $\textbf{BiMix}$'s high accuracy in loss extrapolation (mean relative error < 0.2%) and its generalization to unseen mixtures (R${}^{2}$ > 0.97). Optimization of domain proportions yields superior model performance compared to existing methods. Furthermore, we establish entropy-based measures as efficient proxies for data mixing, offering a computationally lightweight strategy. Our work contributes both theoretical insights into data mixing dynamics and practical tools for enhancing LLM training efficiency, paving the way for more effective scaling strategies in language model development.
Authors: Jiayu Chen, Bhargav Ganguly, Tian Lan, Vaneet Aggarwal
Abstract: Skills are effective temporal abstractions established for sequential decision making, which enable efficient hierarchical learning for long-horizon tasks and facilitate multi-task learning through their transferability. Despite extensive research, research gaps remain in multi-agent scenarios, particularly for automatically extracting subgroup coordination patterns in a multi-agent task. In this case, we propose two novel auto-encoder schemes: VO-MASD-3D and VO-MASD-Hier, to simultaneously capture subgroup- and temporal-level abstractions and form multi-agent skills, which firstly solves the aforementioned challenge. An essential algorithm component of these schemes is a dynamic grouping function that can automatically detect latent subgroups based on agent interactions in a task. Our method can be applied to offline multi-task data, and the discovered subgroup skills can be transferred across relevant tasks without retraining. Empirical evaluations on StarCraft tasks indicate that our approach significantly outperforms existing hierarchical multi-agent reinforcement learning (MARL) methods. Moreover, skills discovered using our method can effectively reduce the learning difficulty in MARL scenarios with delayed and sparse reward signals.
Authors: Shen Yuan, Haotian Liu, Hongteng Xu
Abstract: While following different technical routes, both low-rank and orthogonal adaptation techniques can efficiently adapt large-scale pre-training models in specific tasks or domains based on a small piece of trainable parameters. In this study, we bridge the gap between these two techniques, proposing a simple but effective adaptation method based on Householder reflections. Given a pre-trained model, our method fine-tunes its layers by multiplying each frozen weight matrix with an orthogonal matrix constructed by a chain of learnable Householder reflections (HRs). This HR-based orthogonal fine-tuning is equivalent to an adaptive low-rank adaptation. Moreover, we show that the orthogonality of the reflection planes corresponding to the HRs impacts the model capacity and regularity. The analysis motivates us to regularize the orthogonality of the HRs, leading to different implementations of the proposed Householder reflection adaptation (HRA) method. Compared with state-of-the-art methods, HRA achieves superior performance with fewer learnable parameters when adapting large language models and conditional image generators. The code of the experiments is available at \url{https://github.com/DaShenZi721/HRA}, and the method has been merged into the \href{https://github.com/huggingface/peft}{PEFT} package.
URLs: https://github.com/DaShenZi721/HRA, https://github.com/huggingface/peft
Authors: Giuseppe Serra, Florian Buettner
Abstract: Given the ability to model more realistic and dynamic problems, Federated Continual Learning (FCL) has been increasingly investigated recently. A well-known problem encountered in this setting is the so-called catastrophic forgetting, for which the learning model is inclined to focus on more recent tasks while forgetting the previously learned knowledge. The majority of the current approaches in FCL propose generative-based solutions to solve said problem. However, this setting requires multiple training epochs over the data, implying an offline setting where datasets are stored locally and remain unchanged over time. Furthermore, the proposed solutions are tailored for vision tasks solely. To overcome these limitations, we propose a new approach to deal with different modalities in the online scenario where new data arrive in streams of mini-batches that can only be processed once. To solve catastrophic forgetting, we propose an uncertainty-aware memory-based approach. Specifically, we suggest using an estimator based on the Bregman Information (BI) to compute the model's variance at the sample level. Through measures of predictive uncertainty, we retrieve samples with specific characteristics, and - by retraining the model on such samples - we demonstrate the potential of this approach to reduce the forgetting effect in realistic settings while maintaining data confidentiality and competitive communication efficiency compared to state-of-the-art approaches.
Authors: Xi Lin, Yilu Liu, Xiaoyuan Zhang, Fei Liu, Zhenkun Wang, Qingfu Zhang
Abstract: Multi-objective optimization can be found in many real-world applications where some conflicting objectives can not be optimized by a single solution. Existing optimization methods often focus on finding a set of Pareto solutions with different optimal trade-offs among the objectives. However, the required number of solutions to well approximate the whole Pareto optimal set could be exponentially large with respect to the number of objectives, which makes these methods unsuitable for handling many optimization objectives. In this work, instead of finding a dense set of Pareto solutions, we propose a novel Tchebycheff set scalarization method to find a few representative solutions (e.g., 5) to cover a large number of objectives (e.g., $>100$) in a collaborative and complementary manner. In this way, each objective can be well addressed by at least one solution in the small solution set. In addition, we further develop a smooth Tchebycheff set scalarization approach for efficient optimization with good theoretical guarantees. Experimental studies on different problems with many optimization objectives demonstrate the effectiveness of our proposed method.
Authors: Derek Lim, Theo Moe Putterman, Robin Walters, Haggai Maron, Stefanie Jegelka
Abstract: Many algorithms and observed phenomena in deep learning appear to be affected by parameter symmetries -- transformations of neural network parameters that do not change the underlying neural network function. These include linear mode connectivity, model merging, Bayesian neural network inference, metanetworks, and several other characteristics of optimization or loss-landscapes. However, theoretical analysis of the relationship between parameter space symmetries and these phenomena is difficult. In this work, we empirically investigate the impact of neural parameter symmetries by introducing new neural network architectures that have reduced parameter space symmetries. We develop two methods, with some provable guarantees, of modifying standard neural networks to reduce parameter space symmetries. With these new methods, we conduct a comprehensive experimental study consisting of multiple tasks aimed at assessing the effect of removing parameter symmetries. Our experiments reveal several interesting observations on the empirical impact of parameter symmetries; for instance, we observe linear mode connectivity between our networks without alignment of weight spaces, and we find that our networks allow for faster and more effective Bayesian neural network training. Our code is available at https://github.com/cptq/asymmetric-networks
Authors: Arun D. Kulkarni
Abstract: Recently, convolution neural networks (CNNs) have attracted a great deal of attention due to their remarkable performance in various domains, particularly in image and text classification tasks. However, their application to tabular data classification remains underexplored. There are many fields such as bioinformatics, finance, medicine where nonimage data are prevalent. Adaption of CNNs to classify nonimage data remains highly challenging. This paper investigates the efficacy of CNNs for tabular data classification, aiming to bridge the gap between traditional machine learning approaches and deep learning techniques. We propose a novel framework fuzzy convolution neural network (FCNN) tailored specifically for tabular data to capture local patterns within feature vectors. In our approach, we map feature values to fuzzy memberships. The fuzzy membership vectors are converted into images that are used to train the CNN model. The trained CNN model is used to classify unknown feature vectors. To validate our approach, we generated six complex noisy data sets. We used randomly selected seventy percent samples from each data set for training and thirty percent for testing. The data sets were also classified using the state-of-the-art machine learning algorithms such as the decision tree (DT), support vector machine (SVM), fuzzy neural network (FNN), Bayes classifier, and Random Forest (RF). Experimental results demonstrate that our proposed model can effectively learn meaningful representations from tabular data, achieving competitive or superior performance compared to existing methods. Overall, our finding suggests that the proposed FCNN model holds promise as a viable alternative for tabular data classification tasks, offering a fresh prospective and potentially unlocking new opportunities for leveraging deep learning in structured data analysis.
Authors: Zayd Muhammad Kawakibi Zuhri, Muhammad Farid Adilazuarda, Ayu Purwarianti, Alham Fikri Aji
Abstract: Auto-regressive inference of transformers benefit greatly from Key-Value (KV) caching, but can lead to major memory bottlenecks as model size, batch size, and sequence length grow at scale. We introduce Multi-Layer Key-Value (MLKV) sharing, a novel approach extending KV sharing across transformer layers to reduce memory usage beyond what was possible with Multi-Query Attention (MQA) and Grouped-Query Attention (GQA). Evaluations on various NLP benchmarks and inference metrics using uptrained Pythia-160M variants demonstrate that MLKV significantly reduces memory usage with minimal performance loss, reducing KV cache size down to a factor of 6x compared to MQA. These results highlight MLKV's potential for efficient deployment of transformer models at scale. We provide code at https://github.com/zaydzuhri/pythia-mlkv
Authors: Omer Sahin Tas, Royden Wagner
Abstract: Transformer-based models generate hidden states that are difficult to interpret. In this work, we aim to interpret these hidden states and control them at inference, with a focus on motion forecasting. We leverage the phenomenon of neural collapse and use linear probes to measure interpretable features in hidden states. Our experiments reveal meaningful directions and distances between hidden states of opposing features, which we use to fit control vectors for activation steering. Consequently, our method enables controlling transformer-based motion forecasting models with interpretable features, providing a unique interface to interact with and understand these models. Our implementation is available at https://github.com/kit-mrt/future-motion
Authors: Hui Liu, Wenya Wang, Hao Sun, Chris Xing Tian, Chenqi Kong, Xin Dong, Haoliang Li
Abstract: Large Language Models (LLMs) have demonstrated impressive in-context learning (ICL) capabilities from few-shot demonstration exemplars. While recent learning-based demonstration selection methods have proven beneficial to ICL by choosing more useful exemplars, their underlying mechanisms are opaque, hindering efforts to address limitations such as high training costs and poor generalization across tasks. These methods generally assume the selection process captures similarities between the exemplar and the target instance, however, it remains unknown what kinds of similarities are captured and vital to performing ICL. To dive into this question, we analyze the working mechanisms of the learning-based demonstration selection methods and empirically identify two important factors related to similarity measurement: 1) The ability to integrate different levels of task-agnostic text similarities between the input of exemplars and test cases enhances generalization power across different tasks. 2) Incorporating task-specific labels when measuring the similarities significantly improves the performance on each specific task. We validate these two findings through extensive quantitative and qualitative analyses across ten datasets and various LLMs. Based on our findings, we introduce two effective yet simplified exemplar selection methods catering to task-agnostic and task-specific demands, eliminating the costly LLM inference overhead.
Authors: Tianle Li, Wei-Lin Chiang, Evan Frick, Lisa Dunlap, Tianhao Wu, Banghua Zhu, Joseph E. Gonzalez, Ion Stoica
Abstract: The rapid evolution of Large Language Models (LLMs) has outpaced the development of model evaluation, highlighting the need for continuous curation of new, challenging benchmarks. However, manual curation of high-quality, human-aligned benchmarks is expensive and time-consuming. To address this, we introduce BenchBuilder, an automated pipeline that leverages LLMs to curate high-quality, open-ended prompts from large, crowd-sourced datasets, enabling continuous benchmark updates without human in the loop. We apply BenchBuilder to datasets such as Chatbot Arena and WildChat-1M, extracting challenging prompts and utilizing LLM-as-a-Judge for automatic model evaluation. To validate benchmark quality, we propose new metrics to measure a benchmark's alignment with human preferences and ability to separate models. We release Arena-Hard-Auto, a benchmark consisting 500 challenging prompts curated by BenchBuilder. Arena-Hard-Auto provides 3x higher separation of model performances compared to MT-Bench and achieves 98.6% correlation with human preference rankings, all at a cost of $20. Our work sets a new framework for the scalable curation of automated benchmarks from extensive data.
Authors: Ding Wang
Abstract: In recent years, gas recognition technology has received considerable attention. Nevertheless, the gas recognition area has faced obstacles in implementing deep learning-based recognition solutions due to the absence of standardized protocols. To tackle this problem, we suggest a new GRU. Compared to other models, GRU obtains a higher identification accuracy.
Authors: Dibyajyoti Chakraborty, Seung Whan Chung, Troy Arcomano, Romit Maulik
Abstract: Forecasting high-dimensional dynamical systems is a fundamental challenge in various fields, such as geosciences and engineering. Neural Ordinary Differential Equations (NODEs), which combine the power of neural networks and numerical solvers, have emerged as a promising algorithm for forecasting complex nonlinear dynamical systems. However, classical techniques used for NODE training are ineffective for learning chaotic dynamical systems. In this work, we propose a novel NODE-training approach that allows for robust learning of chaotic dynamical systems. Our method addresses the challenges of non-convexity and exploding gradients associated with underlying chaotic dynamics. Training data trajectories from such systems are split into multiple, non-overlapping time windows. In addition to the deviation from the training data, the optimization loss term further penalizes the discontinuities of the predicted trajectory between the time windows. The window size is selected based on the fastest Lyapunov time scale of the system. Multi-step penalty(MP) method is first demonstrated on Lorenz equation, to illustrate how it improves the loss landscape and thereby accelerates the optimization convergence. MP method can optimize chaotic systems in a manner similar to least-squares shadowing with significantly lower computational costs. Our proposed algorithm, denoted the Multistep Penalty NODE, is applied to chaotic systems such as the Kuramoto-Sivashinsky equation, the two-dimensional Kolmogorov flow, and ERA5 reanalysis data for the atmosphere. It is observed that MP-NODE provide viable performance for such chaotic systems, not only for short-term trajectory predictions but also for invariant statistics that are hallmarks of the chaotic nature of these dynamics.
Authors: Fanzeng Xia, Hao Liu, Yisong Yue, Tongxin Li
Abstract: In-context decision-making is an important capability of artificial general intelligence, which Large Language Models (LLMs) have effectively demonstrated in various scenarios. However, LLMs often face challenges when dealing with numerical contexts, and limited attention has been paid to evaluating their performance through preference feedback generated by the environment. This paper is the first to investigate the performance of LLMs as decision-makers in the context of Dueling Bandits (DB). We compare GPT-3.5 Turbo, GPT-4, GPT-4 Turbo, Llama 3.1, and o1-preview against eight well-established DB algorithms. Our results reveal that LLMs, particularly GPT-4 Turbo, quickly identify the Condorcet winner, thus outperforming existing state-of-the-art algorithms in terms of weak regret. Nevertheless, LLMs struggle to converge even when explicitly prompted to do so and are sensitive to prompt variations. To overcome these issues, we introduce a hybrid algorithm: LLM-Enhanced Adaptive Dueling (LEAD), which takes advantage of both in-context decision-making capabilities of LLMs and theoretical guarantees inherited from classic DB algorithms. We show that LEAD has theoretical guarantees on both weak and strong regret and validate its robustness even with noisy and adversarial prompts. The design of such an algorithm sheds light on how to enhance trustworthiness for LLMs used in decision-making tasks where performance robustness matters.
Authors: Christian Schilling, Martin Zimmermann
Abstract: A control system consists of a plant component and a controller which periodically computes a control input for the plant. We consider systems where the controller is implemented by a feedforward neural network with ReLU activations. The reachability problem asks, given a set of initial states, whether a set of target states can be reached. We show that this problem is undecidable even for trivial plants and fixed-depth neural networks with three inputs and outputs. We also show that the problem becomes semi-decidable when the plant as well as the input and target sets are given by automata over infinite words.
Authors: Bidur Khanal, Tianhong Dai, Binod Bhattarai, Cristian Linte
Abstract: The robustness of supervised deep learning-based medical image classification is significantly undermined by label noise. Although several methods have been proposed to enhance classification performance in the presence of noisy labels, they face some challenges: 1) a struggle with class-imbalanced datasets, leading to the frequent overlooking of minority classes as noisy samples; 2) a singular focus on maximizing performance using noisy datasets, without incorporating experts-in-the-loop for actively cleaning the noisy labels. To mitigate these challenges, we propose a two-phase approach that combines Learning with Noisy Labels (LNL) and active learning. This approach not only improves the robustness of medical image classification in the presence of noisy labels, but also iteratively improves the quality of the dataset by relabeling the important incorrect labels, under a limited annotation budget. Furthermore, we introduce a novel Variance of Gradients approach in LNL phase, which complements the loss-based sample selection by also sampling under-represented samples. Using two imbalanced noisy medical classification datasets, we demonstrate that that our proposed technique is superior to its predecessors at handling class imbalance by not misidentifying clean samples from minority classes as mostly noisy samples.
Authors: Junho Song, Jong-Hwan Jang, Byeong Tak Lee, DongGyun Hong, Joon-myoung Kwon, Yong-Yeon Jo
Abstract: Using foundation models enhanced by self-supervised learning (SSL) methods presents an innovative approach to electrocardiogram (ECG) analysis, which is crucial for cardiac health monitoring and diagnosis. This study comprehensively evaluates foundation models for ECGs, leveraging SSL methods, including generative and contrastive learning, on a vast dataset comprising approximately 1.3 million ECG samples. By integrating these methods with consideration of the unique characteristics of ECGs, we developed a Hybrid Learning (HL) for foundation models that improve the precision and reliability of cardiac diagnostics. The HL-based foundation model adeptly captures the intricate details of ECGs, enhancing diagnostic capability. The results underscore the considerable potential of SSL-enhanced foundation models in clinical settings, setting the stage for future research into their scalable applications across a broader range of medical diagnostics. This work sets a new standard in the ECG field, emphasizing the transformative influence of tailored, data-driven model training on the effectiveness and accuracy of medical diagnostics.
Authors: Mingkai Chen, Taowen Wang, Shihui Cao, James Chenhao Liang, Chuan Liu, Chunshu Wu, Qifan Wang, Ying Nian Wu, Michael Huang, Chuang Ren, Ang Li, Tong Geng, Dongfang Liu
Abstract: Controlled fusion energy is deemed pivotal for the advancement of human civilization. In this study, we introduce $\textbf{LPI-LLM}$, a novel integration of Large Language Models (LLMs) with classical reservoir computing paradigms tailored to address a critical challenge, Laser-Plasma Instabilities ($\texttt{LPI}$), in Inertial Confinement Fusion ($\texttt{ICF}$). Our approach offers several key contributions: Firstly, we propose the $\textit{LLM-anchored Reservoir}$, augmented with a $\textit{Fusion-specific Prompt}$, enabling accurate forecasting of $\texttt{LPI}$-generated-hot electron dynamics during implosion. Secondly, we develop $\textit{Signal-Digesting Channels}$ to temporally and spatially describe the driver laser intensity across time, capturing the unique characteristics of $\texttt{ICF}$ inputs. Lastly, we design the $\textit{Confidence Scanner}$ to quantify the confidence level in forecasting, providing valuable insights for domain experts to design the $\texttt{ICF}$ process. Extensive experiments demonstrate the superior performance of our method, achieving 1.90 CAE, 0.14 $\texttt{top-1}$ MAE, and 0.11 $\texttt{top-5}$ MAE in predicting Hard X-ray ($\texttt{HXR}$) energies emitted by the hot electrons in $\texttt{ICF}$ implosions, which presents state-of-the-art comparisons against concurrent best systems. Additionally, we present $\textbf{LPI4AI}$, the first $\texttt{LPI}$ benchmark based on physical experiments, aimed at fostering novel ideas in $\texttt{LPI}$ research and enhancing the utility of LLMs in scientific exploration. Overall, our work strives to forge an innovative synergy between AI and $\texttt{ICF}$ for advancing fusion energy.
Authors: Pantelis Vafidis, Aman Bhargava, Antonio Rangel
Abstract: Intelligent perception and interaction with the world hinges on internal representations that capture its underlying structure ("disentangled" or "abstract" representations). Disentangled representations serve as world models, isolating latent factors of variation in the world along orthogonal directions, thus facilitating feature-based generalization. We provide experimental and theoretical results guaranteeing the emergence of disentangled representations in agents that optimally solve multi-task evidence aggregation classification tasks, canonical in the cognitive neuroscience literature. The key conceptual finding is that, by producing accurate multi-task classification estimates, a system implicitly represents a set of coordinates specifying a disentangled representation of the underlying latent state of the data it receives. The theory provides conditions for the emergence of these representations in terms of noise, number of tasks, and evidence aggregation time. We experimentally validate these predictions in RNNs trained on multi-task classification, which learn disentangled representations in the form of continuous attractors, leading to zero-shot out-of-distribution (OOD) generalization in predicting latent factors. We demonstrate the robustness of our framework across autoregressive architectures, decision boundary geometries and in tasks requiring classification confidence estimation. We find that transformers are particularly suited for disentangling representations, which might explain their unique world understanding abilities. Overall, our framework puts forth parallel processing as a general principle for the formation of cognitive maps that capture the structure of the world in both biological and artificial systems, and helps explain why ANNs often arrive at human-interpretable concepts, and how they both may acquire exceptional zero-shot generalization capabilities.
Authors: Yingyu Liang, Zhenmei Shi, Zhao Song, Yufa Zhou
Abstract: Cross-attention has become a fundamental module nowadays in many important artificial intelligence applications, e.g., retrieval-augmented generation (RAG), system prompt, guided stable diffusion, and many more. Ensuring cross-attention privacy is crucial and urgently needed because its key and value matrices may contain sensitive information about model providers and their users. In this work, we design a novel differential privacy (DP) data structure to address the privacy security of cross-attention with a theoretical guarantee. In detail, let $n$ be the input token length of system prompt/RAG data, $d$ be the feature dimension, $0 < \alpha \le 1$ be the relative error parameter, $R$ be the maximum value of the query and key matrices, $R_w$ be the maximum value of the value matrix, and $r,s,\epsilon_s$ be parameters of polynomial kernel methods. Then, our data structure requires $\widetilde{O}(ndr^2)$ memory consumption with $\widetilde{O}(nr^2)$ initialization time complexity and $\widetilde{O}(\alpha^{-1} r^2)$ query time complexity for a single token query. In addition, our data structure can guarantee that the process of answering user query satisfies $(\epsilon, \delta)$-DP with $\widetilde{O}(n^{-1} \epsilon^{-1} \alpha^{-1/2} R^{2s} R_w r^2)$ additive error and $n^{-1} (\alpha + \epsilon_s)$ relative error between our output and the true answer. Furthermore, our result is robust to adaptive queries in which users can intentionally attack the cross-attention system. To our knowledge, this is the first work to provide DP for cross-attention and is promising to inspire more privacy algorithm design in large generative models (LGMs).
Authors: Byeong Tak Lee, Joon-myoung Kwon, Yong-Yeon Jo
Abstract: Domain generalization aim to train models to effectively perform on samples that are unseen and outside of the distribution. Adversarial data augmentation (ADA) is a widely used technique in domain generalization. It enhances the model robustness by including synthetic samples designed to simulate potential unseen scenarios into the training datasets, which is then used to train the model. However, in time series data, traditional ADA approaches often fail to address distribution shifts related to temporal characteristics. To address this limitation, we propose Temporal Adversarial Data Augmentation (TADA) for time series data, which incorporate time warping into ADA. Although time warping is inherently non-differentiable, ADA relies on generating samples through backpropagation. We resolve this issue by leveraging the duality between phase shifts in the frequency domain and time shifts in the time domain, thereby making the process differentiable. Our evaluations across various time series datasets demonstrate that TADA outperforms existing methods for domain generalization. In addition, using distribution visualization, we confirmed that the distribution shifts induced by TADA are clearly different from those induced by ADA, and together, they effectively simulate real-world distribution shifts.
Authors: Xin Chen, Sam Toyer, Florian Shkurti
Abstract: Spurious correlations in a reward model's training data can prevent Reinforcement Learning from Human Feedback (RLHF) from identifying the desired goal and induce unwanted behaviors. This paper shows that offline RLHF is susceptible to reward confusion, especially in the presence of spurious correlations in offline data. We create a benchmark to study this problem and propose a method that can significantly reduce reward confusion by leveraging transitivity of preferences while building a global preference chain with active learning.
Authors: Zhengbo Wang, Jian Liang, Ran He, Zilei Wang, Tieniu Tan
Abstract: Low-rank adaptation, also known as LoRA, has emerged as a prominent method for parameter-efficient fine-tuning of foundation models. Despite its computational efficiency, LoRA still yields inferior performance compared to full fine-tuning. In this paper, we first uncover a fundamental connection between the optimization processes of LoRA and full fine-tuning: using LoRA for optimization is mathematically equivalent to full fine-tuning using a low-rank gradient for parameter updates. And this low-rank gradient can be expressed in terms of the gradients of the two low-rank matrices in LoRA. Leveraging this insight, we introduce LoRA-Pro, a method that enhances LoRA's performance by strategically adjusting the gradients of these low-rank matrices. This adjustment allows the low-rank gradient to more accurately approximate the full fine-tuning gradient, thereby narrowing the performance gap between LoRA and full fine-tuning. Furthermore, we theoretically derive the optimal solutions for adjusting the gradients of the low-rank matrices, applying them during fine-tuning in LoRA-Pro. We conduct extensive experiments across natural language understanding, dialogue generation, mathematical reasoning, code generation, and image classification tasks, demonstrating that LoRA-Pro substantially improves LoRA's performance, effectively narrowing the gap with full fine-tuning. Code is publicly available at \url{https://github.com/mrflogs/LoRA-Pro}.
Authors: Khemraj Shukla, Zongren Zou, Chi Hin Chan, Additi Pandey, Zhicheng Wang, George Em Karniadakis
Abstract: Multiphysics problems that are characterized by complex interactions among fluid dynamics, heat transfer, structural mechanics, and electromagnetics, are inherently challenging due to their coupled nature. While experimental data on certain state variables may be available, integrating these data with numerical solvers remains a significant challenge. Physics-informed neural networks (PINNs) have shown promising results in various engineering disciplines, particularly in handling noisy data and solving inverse problems in partial differential equations (PDEs). However, their effectiveness in forecasting nonlinear phenomena in multiphysics regimes, particularly involving turbulence, is yet to be fully established. This study introduces NeuroSEM, a hybrid framework integrating PINNs with the high-fidelity Spectral Element Method (SEM) solver, Nektar++. NeuroSEM leverages the strengths of both PINNs and SEM, providing robust solutions for multiphysics problems. PINNs are trained to assimilate data and model physical phenomena in specific subdomains, which are then integrated into the Nektar++ solver. We demonstrate the efficiency and accuracy of NeuroSEM for thermal convection in cavity flow and flow past a cylinder. We applied NeuroSEM to the Rayleigh-B\'enard convection system, including cases with missing thermal boundary conditions and noisy datasets, and to real particle image velocimetry (PIV) data to capture flow patterns characterized by horseshoe vortical structures. The framework's plug-and-play nature facilitates its extension to other multiphysics or multiscale problems. Furthermore, NeuroSEM is optimized for efficient execution on emerging integrated GPU-CPU architectures. This hybrid approach enhances the accuracy and efficiency of simulations, making it a powerful tool for tackling complex engineering challenges in various scientific domains.
Authors: Marius Huber, Sara Kalisnik, Patrick Schnider
Abstract: We present AuToMATo, a novel clustering algorithm based on persistent homology. While AuToMATo is not parameter-free per se, we provide default choices for its parameters that make it into an out-of-the-box clustering algorithm that performs well across the board. AuToMATo combines the existing ToMATo clustering algorithm with a bootstrapping procedure in order to separate significant peaks of an estimated density function from non-significant ones. We perform a thorough comparison of AuToMATo (with its parameters fixed to their defaults) against many other state-of-the-art clustering algorithms. We find not only that AuToMATo compares favorably against parameter-free clustering algorithms, but in many instances also significantly outperforms even the best selection of parameters for other algorithms. AuToMATo is motivated by applications in topological data analysis, in particular the Mapper algorithm, where it is desirable to work with a clustering algorithm that does not need tuning of its parameters. Indeed, we provide evidence that AuToMATo performs well when used with Mapper. Finally, we provide an open-source implementation of AuToMATo in Python that is fully compatible with the standard scikit-learn architecture.
Authors: Subrat Prasad Panda, Blaise Genest, Arvind Easwaran, Ponnuthurai Nagaratnam Suganthan
Abstract: Decision Trees (DTs) constitute one of the major highly non-linear AI models, valued, e.g., for their efficiency on tabular data. Learning accurate DTs is, however, complicated, especially for oblique DTs, and does take a significant training time. Further, DTs suffer from overfitting, e.g., they proverbially "do not generalize" in regression tasks. Recently, some works proposed ways to make (oblique) DTs differentiable. This enables highly efficient gradient-descent algorithms to be used to learn DTs. It also enables generalizing capabilities by learning regressors at the leaves simultaneously with the decisions in the tree. Prior approaches to making DTs differentiable rely either on probabilistic approximations at the tree's internal nodes (soft DTs) or on approximations in gradient computation at the internal node (quantized gradient descent). In this work, we propose DTSemNet, a novel semantically equivalent and invertible encoding for (hard, oblique) DTs as Neural Networks (NNs), that uses standard vanilla gradient descent. Experiments across various classification and regression benchmarks show that oblique DTs learned using DTSemNet are more accurate than oblique DTs of similar size learned using state-of-the-art techniques. Further, DT training time is significantly reduced. We also experimentally demonstrate that DTSemNet can learn DT policies as efficiently as NN policies in the Reinforcement Learning (RL) setup with physical inputs (dimensions $\leq32$). The code is available at https://github.com/CPS-research-group/dtsemnet.
Authors: Yingyu Liang, Zhizhou Sha, Zhenmei Shi, Zhao Song, Yufa Zhou
Abstract: The computational complexity of the self-attention mechanism in popular transformer architectures poses significant challenges for training and inference, and becomes the bottleneck for long inputs. Is it possible to significantly reduce the quadratic time complexity of computing the gradients in multi-layer transformer models? This paper proves that a novel fast approximation method can calculate the gradients in almost linear time $n^{1+o(1)}$ where $n$ is the input sequence length, while it maintains a polynomially small approximation error $1 / \mathrm{poly}(n)$ across the entire model. Our theory holds for general loss functions and when the multi-layer transformer model contains many practical sub-modules, such as residual connection, casual mask, and multi-head attention. By improving the efficiency of gradient computation, we hope that this work will facilitate more effective training and deployment of long-context language models based on our theoretical results.
Authors: Ofir Schlisselberg, Ido Cohen, Tal Lancewicki, Yishay Mansour
Abstract: In this paper, we investigate a variant of the classical stochastic Multi-armed Bandit (MAB) problem, where the payoff received by an agent (either cost or reward) is both delayed, and directly corresponds to the magnitude of the delay. This setting models faithfully many real world scenarios such as the time it takes for a data packet to traverse a network given a choice of route (where delay serves as the agent's cost); or a user's time spent on a web page given a choice of content (where delay serves as the agent's reward). Our main contributions are tight upper and lower bounds for both the cost and reward settings. For the case that delays serve as costs, which we are the first to consider, we prove optimal regret that scales as $\sum_{i:\Delta_i > 0}\frac{\log T}{\Delta_i} + d^*$, where $T$ is the maximal number of steps, $\Delta_i$ are the sub-optimality gaps and $d^*$ is the minimal expected delay amongst arms. For the case that delays serves as rewards, we show optimal regret of $\sum_{i:\Delta_i > 0}\frac{\log T}{\Delta_i} + \bar{d}$, where $\bar d$ is the second maximal expected delay. These improve over the regret in the general delay-dependent payoff setting, which scales as $\sum_{i:\Delta_i > 0}\frac{\log T}{\Delta_i} + D$, where $D$ is the maximum possible delay. Our regret bounds highlight the difference between the cost and reward scenarios, showing that the improvement in the cost scenario is more significant than for the reward. Finally, we accompany our theoretical results with an empirical evaluation.
Authors: Osama Ahmad, Zubair Khalid
Abstract: This paper focuses on spatiotemporal (ST) traffic prediction using graph neural networks. Given that ST data consists of non-stationary and complex time events, interpreting and predicting such trends is comparatively complicated. Representation of ST data in modes helps us to infer behavior and assess the impact of noise on prediction applications. We propose a framework that decomposes ST data into modes using the variational mode decomposition (VMD) method, which is then fed into the neural network for forecasting future states. This hybrid approach is known as a variational mode graph convolutional network (VMGCN). Instead of exhaustively searching for the number of modes, they are determined using the reconstruction loss from the real-time application data. We also study the significance of each mode and the impact of bandwidth constraints on different horizon predictions in traffic flow data. We evaluate the performance of our proposed network on the LargeST dataset for both short and long-term predictions. Our framework yields better results compared to state-of-the-art methods.
Authors: Ziang Liu, Junjie Xu, Xingjiao Wu, Jing Yang, Liang He
Abstract: Preference-Based reinforcement learning (PBRL) learns directly from the preferences of human teachers regarding agent behaviors without needing meticulously designed reward functions. However, existing PBRL methods often learn primarily from explicit preferences, neglecting the possibility that teachers may choose equal preferences. This neglect may hinder the understanding of the agent regarding the task perspective of the teacher, leading to the loss of important information. To address this issue, we introduce the Equal Preference Learning Task, which optimizes the neural network by promoting similar reward predictions when the behaviors of two agents are labeled as equal preferences. Building on this task, we propose a novel PBRL method, Multi-Type Preference Learning (MTPL), which allows simultaneous learning from equal preferences while leveraging existing methods for learning from explicit preferences. To validate our approach, we design experiments applying MTPL to four existing state-of-the-art baselines across ten locomotion and robotic manipulation tasks in the DeepMind Control Suite. The experimental results indicate that simultaneous learning from both equal and explicit preferences enables the PBRL method to more comprehensively understand the feedback from teachers, thereby enhancing feedback efficiency. Project page: \url{https://github.com/FeiCuiLengMMbb/paper_MTPL}
Authors: Hrishikesh Patel, Ruihong Qiu, Adam Irwin, Shazia Sadiq, Sen Wang
Abstract: Irregular time series, where data points are recorded at uneven intervals, are prevalent in healthcare settings, such as emergency wards where vital signs and laboratory results are captured at varying times. This variability, which reflects critical fluctuations in patient health, is essential for informed clinical decision-making. Existing self-supervised learning research on irregular time series often relies on generic pretext tasks like forecasting, which may not fully utilise the signal provided by irregular time series. There is a significant need for specialised pretext tasks designed for the characteristics of irregular time series to enhance model performance and robustness, especially in scenarios with limited data availability. This paper proposes a novel pretraining framework, EMIT, an event-based masking for irregular time series. EMIT focuses on masking-based reconstruction in the latent space, selecting masking points based on the rate of change in the data. This method preserves the natural variability and timing of measurements while enhancing the model's ability to process irregular intervals without losing essential information. Extensive experiments on the MIMIC-III and PhysioNet Challenge datasets demonstrate the superior performance of our event-based masking strategy. The code has been released at https://github.com/hrishi-ds/EMIT.
Authors: Shengsheng Lin, Weiwei Lin, Xinyi Hu, Wentai Wu, Ruichao Mo, Haocheng Zhong
Abstract: The stable periodic patterns present in time series data serve as the foundation for conducting long-horizon forecasts. In this paper, we pioneer the exploration of explicitly modeling this periodicity to enhance the performance of models in long-term time series forecasting (LTSF) tasks. Specifically, we introduce the Residual Cycle Forecasting (RCF) technique, which utilizes learnable recurrent cycles to model the inherent periodic patterns within sequences, and then performs predictions on the residual components of the modeled cycles. Combining RCF with a Linear layer or a shallow MLP forms the simple yet powerful method proposed in this paper, called CycleNet. CycleNet achieves state-of-the-art prediction accuracy in multiple domains including electricity, weather, and energy, while offering significant efficiency advantages by reducing over 90% of the required parameter quantity. Furthermore, as a novel plug-and-play technique, the RCF can also significantly improve the prediction accuracy of existing models, including PatchTST and iTransformer. The source code is available at: https://github.com/ACAT-SCUT/CycleNet.
Authors: Hongjun Wang, Jiyuan Chen, Tong Pan, Zheng Dong, Lingyu Zhang, Renhe Jiang, Xuan Song
Abstract: Traffic forecasting is a cornerstone of smart city management, enabling efficient resource allocation and transportation planning. Deep learning, with its ability to capture complex nonlinear patterns in spatiotemporal (ST) data, has emerged as a powerful tool for traffic forecasting. While graph neural networks (GCNs) and transformer-based models have shown promise, their computational demands often hinder their application to real-world road networks, particularly those with large-scale spatiotemporal interactions. To address these challenges, we propose a novel spatiotemporal graph transformer (STGformer) architecture. STGformer effectively balances the strengths of GCNs and Transformers, enabling efficient modeling of both global and local traffic patterns while maintaining a manageable computational footprint. Unlike traditional approaches that require multiple attention layers, STG attention block captures high-order spatiotemporal interactions in a single layer, significantly reducing computational cost. In particular, STGformer achieves a 100x speedup and a 99.8\% reduction in GPU memory usage compared to STAEformer during batch inference on a California road graph with 8,600 sensors. We evaluate STGformer on the LargeST benchmark and demonstrate its superiority over state-of-the-art Transformer-based methods such as PDFormer and STAEformer, which underline STGformer's potential to revolutionize traffic forecasting by overcoming the computational and memory limitations of existing approaches, making it a promising foundation for future spatiotemporal modeling tasks.
Authors: Cooper Lorsung, Amir Barati Farimani
Abstract: Solving Partial Differential Equations (PDEs) is ubiquitous in science and engineering. Computational complexity and difficulty in writing numerical solvers has motivated the development of machine learning techniques to generate solutions quickly. Many existing methods are purely data driven, relying solely on numerical solution fields, rather than known system information such as boundary conditions and governing equations. However, the recent rise in popularity of Large Language Models (LLMs) has enabled easy integration of text in multimodal machine learning models. In this work, we use pretrained LLMs to integrate various amounts known system information into PDE learning. Our multimodal approach significantly outperforms our baseline model, FactFormer, in both next-step prediction and autoregressive rollout performance on the 2D Heat, Burgers, Navier-Stokes, and Shallow Water equations. Further analysis shows that pretrained LLMs provide highly structured latent space that is consistent with the amount of system information provided through text.
Authors: Erica Zhang, Fangzhao Zhang, Mert Pilanci
Abstract: Active learning methods aim to improve sample complexity in machine learning. In this work, we investigate an active learning scheme via a novel gradient-free cutting-plane training method for ReLU networks of arbitrary depth. We demonstrate, for the first time, that cutting-plane algorithms, traditionally used in linear models, can be extended to deep neural networks despite their nonconvexity and nonlinear decision boundaries. Our results demonstrate that these methods provide a promising alternative to the commonly employed gradient-based optimization techniques in large-scale neural networks. Moreover, this training method induces the first deep active learning scheme known to achieve convergence guarantees. We exemplify the effectiveness of our proposed active learning method against popular deep active learning baselines via both synthetic data experiments and sentimental classification task on real datasets.
Authors: Ulyana Piterbarg, Lerrel Pinto, Rob Fergus
Abstract: Software engineers mainly write code by editing existing programs. In contrast, large language models (LLMs) autoregressively synthesize programs in a single pass. One explanation for this is the scarcity of open-sourced edit data. While high-quality instruction data for code synthesis is already scarce, high-quality edit data is even scarcer. To fill this gap, we develop a synthetic data generation algorithm called LintSeq. This algorithm refactors existing code into a sequence of code edits by using a linter to procedurally sample across the error-free insertions that can be used to sequentially write programs. It outputs edit sequences as text strings consisting of consecutive program diffs. To test LintSeq, we use it to refactor a dataset of instruction + program pairs into instruction + program-diff-sequence tuples. Then, we instruction finetune a series of smaller LLMs ranging from 2.6B to 14B parameters on both the re-factored and original versions of this dataset, comparing zero-shot performance on code synthesis benchmarks. We show that during repeated sampling, edit sequence finetuned models produce more diverse programs than baselines. This results in better inference-time scaling for benchmark coverage as a function of samples, i.e. the fraction of problems "pass@k" solved by any attempt given "k" tries. For example, on HumanEval pass@50, small LLMs finetuned on synthetic edit sequences are competitive with GPT-4 and outperform models finetuned on the baseline dataset by +20% (+/-3%) in absolute score. Finally, we also pretrain our own tiny LMs for code understanding. We show that finetuning tiny models on synthetic code edits results in state-of-the-art code synthesis for the on-device model class. Our 150M parameter edit sequence LM matches or outperforms code models with twice as many parameters, both with and without repeated sampling, including Codex and AlphaCode.
Authors: Isaac Reid, Kumar Avinava Dubey, Deepali Jain, Will Whitney, Amr Ahmed, Joshua Ainslie, Alex Bewley, Mithun Jacob, Aranyak Mehta, David Rendleman, Connor Schenck, Richard E. Turner, Ren\'e Wagner, Adrian Weller, Krzysztof Choromanski
Abstract: When training transformers on graph-structured data, incorporating information about the underlying topology is crucial for good performance. Topological masking, a type of relative position encoding, achieves this by upweighting or downweighting attention depending on the relationship between the query and keys in a graph. In this paper, we propose to parameterise topological masks as a learnable function of a weighted adjacency matrix -- a novel, flexible approach which incorporates a strong structural inductive bias. By approximating this mask with graph random features (for which we prove the first known concentration bounds), we show how this can be made fully compatible with linear attention, preserving $\mathcal{O}(N)$ time and space complexity with respect to the number of input tokens. The fastest previous alternative was $\mathcal{O}(N \log N)$ and only suitable for specific graphs. Our efficient masking algorithms provide strong performance gains for tasks on image and point cloud data, including with $>30$k nodes.
Authors: Theo Putterman, Derek Lim, Yoav Gelberg, Stefanie Jegelka, Haggai Maron
Abstract: Low-rank adaptations (LoRAs) have revolutionized the finetuning of large foundation models, enabling efficient adaptation even with limited computational resources. The resulting proliferation of LoRAs presents exciting opportunities for applying machine learning techniques that take these low-rank weights themselves as inputs. In this paper, we investigate the potential of Learning on LoRAs (LoL), a paradigm where LoRA weights serve as input to machine learning models. For instance, an LoL model that takes in LoRA weights as inputs could predict the performance of the finetuned model on downstream tasks, detect potentially harmful finetunes, or even generate novel model edits without traditional training methods. We first identify the inherent parameter symmetries of low rank decompositions of weights, which differ significantly from the parameter symmetries of standard neural networks. To efficiently process LoRA weights, we develop several symmetry-aware invariant or equivariant LoL models, using tools such as canonicalization, invariant featurization, and equivariant layers. We finetune thousands of text-to-image diffusion models and language models to collect datasets of LoRAs. In numerical experiments on these datasets, we show that our LoL architectures are capable of processing low rank weight decompositions to predict CLIP score, finetuning data attributes, finetuning data membership, and accuracy on downstream tasks.
Authors: Yongyi Su, Yushu Li, Nanqing Liu, Kui Jia, Xulei Yang, Chuan-Sheng Foo, Xun Xu
Abstract: Test-time adaptation (TTA) updates the model weights during the inference stage using testing data to enhance generalization. However, this practice exposes TTA to adversarial risks. Existing studies have shown that when TTA is updated with crafted adversarial test samples, also known as test-time poisoned data, the performance on benign samples can deteriorate. Nonetheless, the perceived adversarial risk may be overstated if the poisoned data is generated under overly strong assumptions. In this work, we first review realistic assumptions for test-time data poisoning, including white-box versus grey-box attacks, access to benign data, attack budget, and more. We then propose an effective and realistic attack method that better produces poisoned samples without access to benign samples, and derive an effective in-distribution attack objective. We also design two TTA-aware attack objectives. Our benchmarks of existing attack methods reveal that the TTA methods are more robust than previously believed. In addition, we analyze effective defense strategies to help develop adversarially robust TTA methods.
Authors: Francesco Riccardo Crescenzi
Abstract: The unprecedented pace of machine learning research has lead to incredible advances, but also poses hard challenges. At present, the field lacks strong theoretical underpinnings, and many important achievements stem from ad hoc design choices which are hard to justify in principle and whose effectiveness often goes unexplained. Research debt is increasing and many papers are found not to be reproducible. This thesis is a survey that covers some recent work attempting to study machine learning categorically. Category theory is a branch of abstract mathematics that has found successful applications in many fields, both inside and outside mathematics. Acting as a lingua franca of mathematics and science, category theory might be able to give a unifying structure to the field of machine learning. This could solve some of the aforementioned problems. In this work, we mainly focus on the application of category theory to deep learning. Namely, we discuss the use of categorical optics to model gradient-based learning, the use of categorical algebras and integral transforms to link classical computer science to neural networks, the use of functors to link different layers of abstraction and preserve structure, and, finally, the use of string diagrams to provide detailed representations of neural network architectures.
Authors: Zihao Zhou, Rose Yu
Abstract: Large Language Models (LLMs) have gained popularity in time series forecasting, but their potential for anomaly detection remains largely unexplored. Our study investigates whether LLMs can understand and detect anomalies in time series data, focusing on zero-shot and few-shot scenarios. Inspired by conjectures about LLMs' behavior from time series forecasting research, we formulate key hypotheses about LLMs' capabilities in time series anomaly detection. We design and conduct principled experiments to test each of these hypotheses. Our investigation reveals several surprising findings about LLMs for time series: 1. LLMs understand time series better as images rather than as text 2. LLMs did not demonstrate enhanced performance when prompted to engage in explicit reasoning about time series analysis 3. Contrary to common beliefs, LLM's understanding of time series do not stem from their repetition biases or arithmetic abilities 4. LLMs' behaviors and performance in time series analysis vary significantly across different model architectures This study provides the first comprehensive analysis of contemporary LLM capabilities in time series anomaly detection. Our results suggest that while LLMs can understand time series anomalies, many common conjectures based on their reasoning capabilities do not hold. Our code and data are available at `https://github.com/Rose-STL-Lab/AnomLLM/`.
Authors: Xueru Wen, Jie Lou, Yaojie Lu, Hongyu Lin, Xing Yu, Xinyu Lu, Ben He, Xianpei Han, Debing Zhang, Le Sun
Abstract: Reward Models (RMs) are crucial for aligning language models with human preferences. Currently, the evaluation of RMs depends on measuring accuracy against a validation set of manually annotated preference data. Although this method is straightforward and widely adopted, the relationship between RM accuracy and downstream policy performance remains under-explored. In this work, we conduct experiments in a synthetic setting to investigate how differences in RM measured by accuracy translate into gaps in optimized policy performance. Our findings reveal that while there is a weak positive correlation between accuracy and downstream performance, policies optimized towards RMs with similar accuracy can exhibit quite different performance. Moreover, we discover that the way of measuring accuracy significantly impacts its ability to predict the final policy performance. Through the lens of Regressional Goodhart's effect, we identify the existence of exogenous variables impacting the relationship between RM quality measured by accuracy and policy model capability. This underscores the inadequacy of relying solely on accuracy to reflect their impact on policy optimization.
Authors: Nikita Balagansky, Ian Maksimov, Daniil Gavrilov
Abstract: Understanding how features evolve across layers in deep neural networks is a fundamental challenge in mechanistic interpretability, particularly due to polysemanticity and feature superposition. While Sparse Autoencoders (SAEs) have been used to extract interpretable features from individual layers, aligning these features across layers has remained an open problem. In this paper, we introduce SAE Match, a novel, data-free method for aligning SAE features across different layers of a neural network. Our approach involves matching features by minimizing the mean squared error between the folded parameters of SAEs, a technique that incorporates activation thresholds into the encoder and decoder weights to account for differences in feature scales. Through extensive experiments on the Gemma 2 language model, we demonstrate that our method effectively captures feature evolution across layers, improving feature matching quality. We also show that features persist over several layers and that our approach can approximate hidden states across layers. Our work advances the understanding of feature dynamics in neural networks and provides a new tool for mechanistic interpretability studies.
Authors: Xinyu Liao, Aoyang Qin, Jacob Seidman, Junqi Wang, Wei Wang, Paris Perdikaris
Abstract: Most existing generative models are limited to learning a single probability distribution from the training data and cannot generalize to novel distributions for unseen data. An architecture that can generate samples from both trained datasets and unseen probability distributions would mark a significant breakthrough. Recently, score-based generative models have gained considerable attention for their comprehensive mode coverage and high-quality image synthesis, as they effectively learn an operator that maps a probability distribution to its corresponding score function. In this work, we introduce the $\emph{Score Neural Operator}$, which learns the mapping from multiple probability distributions to their score functions within a unified framework. We employ latent space techniques to facilitate the training of score matching, which tends to over-fit in the original image pixel space, thereby enhancing sample generation quality. Our trained Score Neural Operator demonstrates the ability to predict score functions of probability measures beyond the training space and exhibits strong generalization performance in both 2-dimensional Gaussian Mixture Models and 1024-dimensional MNIST double-digit datasets. Importantly, our approach offers significant potential for few-shot learning applications, where a single image from a new distribution can be leveraged to generate multiple distinct images from that distribution.
Authors: Yaming Yang, Dilxat Muhtar, Yelong Shen, Yuefeng Zhan, Jianfeng Liu, Yujing Wang, Hao Sun, Denvy Deng, Feng Sun, Qi Zhang, Weizhu Chen, Yunhai Tong
Abstract: Parameter-efficient fine-tuning (PEFT) has been widely employed for domain adaptation, with LoRA being one of the most prominent methods due to its simplicity and effectiveness. However, in multi-task learning (MTL) scenarios, LoRA tends to obscure the distinction between tasks by projecting sparse high-dimensional features from different tasks into the same dense low-dimensional intrinsic space. This leads to task interference and suboptimal performance for LoRA and its variants. To tackle this challenge, we propose MTL-LoRA, which retains the advantages of low-rank adaptation while significantly enhancing multi-task learning capabilities. MTL-LoRA augments LoRA by incorporating additional task-adaptive parameters that differentiate task-specific information and effectively capture shared knowledge across various tasks within low-dimensional spaces. This approach enables large language models (LLMs) pre-trained on general corpus to adapt to different target task domains with a limited number of trainable parameters. Comprehensive experimental results, including evaluations on public academic benchmarks for natural language understanding, commonsense reasoning, and image-text understanding, as well as real-world industrial text Ads relevance datasets, demonstrate that MTL-LoRA outperforms LoRA and its various variants with comparable or even fewer learnable parameters in multitask learning.
Authors: Jake Fawkes, Nic Fishman, Mel Andrews, Zachary C. Lipton
Abstract: Fairness metrics are a core tool in the fair machine learning literature (FairML), used to determine that ML models are, in some sense, "fair". Real-world data, however, are typically plagued by various measurement biases and other violated assumptions, which can render fairness assessments meaningless. We adapt tools from causal sensitivity analysis to the FairML context, providing a general framework which (1) accommodates effectively any combination of fairness metric and bias that can be posed in the "oblivious setting"; (2) allows researchers to investigate combinations of biases, resulting in non-linear sensitivity; and (3) enables flexible encoding of domain-specific constraints and assumptions. Employing this framework, we analyze the sensitivity of the most common parity metrics under 3 varieties of classifier across 14 canonical fairness datasets. Our analysis reveals the striking fragility of fairness assessments to even minor dataset biases. We show that causal sensitivity analysis provides a powerful and necessary toolkit for gauging the informativeness of parity metric evaluations. Our repository is available here: https://github.com/Jakefawkes/fragile_fair.
Authors: Vithursan Thangarasa, Ganesh Venkatesh, Nish Sinnadurai, Sean Lie
Abstract: Large language models have driven significant progress in natural language processing, but their deployment requires substantial compute and memory resources. As models scale, compression techniques become essential for balancing model quality with computational efficiency. Structured pruning, which removes less critical components of the model, is a promising strategy for reducing complexity. However, one-shot pruning often results in significant quality degradation, particularly in tasks requiring multi-step reasoning. To recover lost quality, supervised fine-tuning (SFT) is commonly applied, but it can lead to catastrophic forgetting by shifting the model's learned data distribution. Therefore, addressing the degradation from both pruning and SFT is essential to preserve the original model's quality. In this work, we propose self-data distilled fine-tuning to address these challenges. Our approach leverages the original, unpruned model to generate a distilled dataset that preserves semantic richness and mitigates catastrophic forgetting by maintaining alignment with the base model's knowledge. Empirically, we demonstrate that self-data distillation consistently outperforms standard SFT, improving average accuracy by up to 8% on the HuggingFace OpenLLM Leaderboard v1. Specifically, when pruning 6 decoder blocks on Llama3.1-8B Instruct (i.e., 32 to 26 layers, reducing the model size from 8.03B to 6.72B parameters), our method retains 91.2% of the original model's accuracy compared to 81.7% with SFT, while reducing real-world FLOPs by 16.30%. Furthermore, our approach scales effectively across datasets, with the quality improving as the dataset size increases.
Authors: Xiaoyu Xia, Ziqi Wang, Ruoxi Sun, Bowen Liu, Ibrahim Khalil, Minhui Xue
Abstract: The right to be forgotten mandates that machine learning models enable the erasure of a data owner's data and information from a trained model. Removing data from the dataset alone is inadequate, as machine learning models can memorize information from the training data, increasing the potential privacy risk to users. To address this, multiple machine unlearning techniques have been developed and deployed. Among them, approximate unlearning is a popular solution, but recent studies report that its unlearning effectiveness is not fully guaranteed. Another approach, exact unlearning, tackles this issue by discarding the data and retraining the model from scratch, but at the cost of considerable computational and memory resources. However, not all devices have the capability to perform such retraining. In numerous machine learning applications, such as edge devices, Internet-of-Things (IoT), mobile devices, and satellites, resources are constrained, posing challenges for deploying existing exact unlearning methods. In this study, we propose a Constraint-aware Adaptive Exact Unlearning System at the network Edge (CAUSE), an approach to enabling exact unlearning on resource-constrained devices. Aiming to minimize the retrain overhead by storing sub-models on the resource-constrained device, CAUSE innovatively applies a Fibonacci-based replacement strategy and updates the number of shards adaptively in the user-based data partition process. To further improve the effectiveness of memory usage, CAUSE leverages the advantage of model pruning to save memory via compression with minimal accuracy sacrifice. The experimental results demonstrate that CAUSE significantly outperforms other representative systems in realizing exact unlearning on the resource-constrained device by 9.23%-80.86%, 66.21%-83.46%, and 5.26%-194.13% in terms of unlearning speed, energy consumption, and accuracy.
Authors: Julius Aka, Johannes Brunnemann, J\"org Eiden, Arne Speerforck, Lars Mikelsons
Abstract: Variational Autoencoders (VAEs) are a powerful framework for learning compact latent representations, while NeuralODEs excel in learning transient system dynamics. This work combines the strengths of both to create fast surrogate models with adjustable complexity. By leveraging the VAE's dimensionality reduction using a non-hierarchical prior, our method adaptively assigns stochastic noise, naturally complementing known NeuralODE training enhancements and enabling probabilistic time series modeling. We show that standard Latent ODEs struggle with dimensionality reduction in systems with time-varying inputs. Our approach mitigates this by continuously propagating variational parameters through time, establishing fixed information channels in latent space. This results in a flexible and robust method that can learn different system complexities, e.g. deep neural networks or linear matrices. Hereby, it enables efficient approximation of the Koopman operator without the need for predefining its dimensionality. As our method balances dimensionality reduction and reconstruction accuracy, we call it Balanced Neural ODE (B-NODE). We demonstrate the effectiveness of this method on academic test cases and apply it to a real-world example of a thermal power plant.
Authors: Tung Nguyen, Qiuyi Zhang, Bangding Yang, Chansoo Lee, Jorg Bornschein, Yingjie Miao, Sagi Perel, Yutian Chen, Xingyou Song
Abstract: Bayesian Optimization is ubiquitous in the field of experimental design and blackbox optimization for improving search efficiency, but has been traditionally restricted to regression models which are only applicable to fixed search spaces and tabular input features. We propose Embed-then-Regress, a paradigm for applying in-context regression over string inputs, through the use of string embedding capabilities of pretrained language models. By expressing all inputs as strings, we are able to perform general-purpose regression for Bayesian Optimization over various domains including synthetic, combinatorial, and hyperparameter optimization, obtaining comparable results to state-of-the-art Gaussian Process-based algorithms. Code can be found at https://github.com/google-research/optformer/tree/main/optformer/embed_then_regress.
URLs: https://github.com/google-research/optformer/tree/main/optformer/embed_then_regress.
Authors: Yun Zhu, Haizhou Shi, Xiaotang Wang, Yongchao Liu, Yaoke Wang, Boci Peng, Chuntao Hong, Siliang Tang
Abstract: Recently, research on Text-Attributed Graphs (TAGs) has gained significant attention due to the prevalence of free-text node features in real-world applications and the advancements in Large Language Models (LLMs) that bolster TAG methodologies. However, current TAG approaches face two primary challenges: (i) Heavy reliance on label information and (ii) Limited cross-domain zero/few-shot transferability. These issues constrain the scaling of both data and model size, owing to high labor costs and scaling laws, complicating the development of graph foundation models with strong transferability. In this work, we propose the GraphCLIP framework to address these challenges by learning graph foundation models with strong cross-domain zero/few-shot transferability through a self-supervised contrastive graph-summary pretraining method. Specifically, we generate and curate large-scale graph-summary pair data with the assistance of LLMs, and introduce a novel graph-summary pretraining method, combined with invariant learning, to enhance graph foundation models with strong cross-domain zero-shot transferability. For few-shot learning, we propose a novel graph prompt tuning technique aligned with our pretraining objective to mitigate catastrophic forgetting and minimize learning costs. Extensive experiments show the superiority of GraphCLIP in both zero-shot and few-shot settings, while evaluations across various downstream tasks confirm the versatility of GraphCLIP. Our code is available at: https://github.com/ZhuYun97/GraphCLIP
Authors: Krzysztof Choromanski, Isaac Reid, Arijit Sehanobish, Avinava Dubey
Abstract: We present the first linear time complexity randomized algorithms for unbiased approximation of the celebrated family of general random walk kernels (RWKs) for sparse graphs. This includes both labelled and unlabelled instances. The previous fastest methods for general RWKs were of cubic time complexity and not applicable to labelled graphs. Our method samples dependent random walks to compute novel graph embeddings in $\mathbb{R}^d$ whose dot product is equal to the true RWK in expectation. It does so without instantiating the direct product graph in memory, meaning we can scale to massive datasets that cannot be stored on a single machine. We derive exponential concentration bounds to prove that our estimator is sharp, and show that the ability to approximate general RWKs (rather than just special cases) unlocks efficient implicit graph kernel learning. Our method is up to $\mathbf{27\times}$ faster than its counterparts for efficient computation on large graphs and scales to graphs $\mathbf{128 \times}$ bigger than largest examples amenable to brute-force computation.
Authors: Bart van Erp, Bert de Vries
Abstract: This paper proposes improvements over earlier work by Nazareth and Blei (2022) for estimating the depth of Bayesian neural networks. Here, we propose a discrete truncated normal distribution over the network depth to independently learn its mean and variance. Posterior distributions are inferred by minimizing the variational free energy, which balances the model complexity and accuracy. Our method improves test accuracy on the spiral data set and reduces the variance in posterior depth estimates.
Authors: Haoyu Tu, Lin Chen, Zuguang Li, Xiaopei Chen, Wen Wu
Abstract: In this paper, we study a vehicle selection problem for federated learning (FL) over vehicular networks. Specifically, we design a mobility-aware vehicular federated learning (MAVFL) scheme in which vehicles drive through a road segment to perform FL. Some vehicles may drive out of the segment which leads to unsuccessful training. In the proposed scheme, the real-time successful training participation ratio is utilized to implement vehicle selection. We conduct the convergence analysis to indicate the influence of vehicle mobility on training loss. Furthermore, we propose a multi-armed bandit-based vehicle selection algorithm to minimize the utility function considering training loss and delay. The simulation results show that compared with baselines, the proposed algorithm can achieve better training performance with approximately 28\% faster convergence.
Authors: Alessio Gravina
Abstract: Graphs are a highly expressive abstraction for modeling entities and their relations, such as molecular structures, social networks, and traffic networks. Deep Graph Networks (DGNs) have emerged as a family of deep learning models that can effectively process and learn such structured information. However, learning effective information propagation patterns within DGNs remains a critical challenge that heavily influences the model capabilities, both in the static domain and in the temporal domain (where features and/or topology evolve). Given this challenge, this thesis investigates the dynamics of information propagation within DGNs for static and dynamic graphs, focusing on their design as dynamical systems. Throughout this work, we provide theoretical and empirical evidence to demonstrate the effectiveness of our proposed architectures in propagating and preserving long-term dependencies between nodes, and in learning complex spatio-temporal patterns from irregular and sparsely sampled dynamic graphs. In summary, this thesis provides a comprehensive exploration of the intersection between graphs, deep learning, and dynamical systems, offering insights and advancements for the field of graph representation learning and paving the way for more effective and versatile graph-based learning models.
Authors: Luofeng Liao, Zuyue Fu, Zhuoran Yang, Yixin Wang, Mladen Kolar, Zhaoran Wang
Abstract: In offline reinforcement learning (RL) an optimal policy is learned solely from a priori collected observational data. However, in observational data, actions are often confounded by unobserved variables. Instrumental variables (IVs), in the context of RL, are the variables whose influence on the state variables is all mediated by the action. When a valid instrument is present, we can recover the confounded transition dynamics through observational data. We study a confounded Markov decision process where the transition dynamics admit an additive nonlinear functional form. Using IVs, we derive a conditional moment restriction through which we can identify transition dynamics based on observational data. We propose a provably efficient IV-aided Value Iteration (IVVI) algorithm based on a primal-dual reformulation of the conditional moment restriction. To our knowledge, this is the first provably efficient algorithm for instrument-aided offline RL.
Authors: Alberto Floris, Luca Frittoli, Diego Carrera, Giacomo Boracchi
Abstract: Deep neural networks require specific layers to process point clouds, as the scattered and irregular location of 3D points prevents the use of conventional convolutional filters. We introduce the composite layer, a flexible and general alternative to the existing convolutional operators that process 3D point clouds. We design our composite layer to extract and compress the spatial information from the 3D coordinates of points and then combine this with the feature vectors. Compared to mainstream point-convolutional layers such as ConvPoint and KPConv, our composite layer guarantees greater flexibility in network design and provides an additional form of regularization. To demonstrate the generality of our composite layers, we define both a convolutional composite layer and an aggregate version that combines spatial information and features in a nonlinear manner, and we use these layers to implement CompositeNets. Our experiments on synthetic and real-world datasets show that, in both classification, segmentation, and anomaly detection, our CompositeNets outperform ConvPoint, which uses the same sequential architecture, and achieve similar results as KPConv, which has a deeper, residual architecture. Moreover, our CompositeNets achieve state-of-the-art performance in anomaly detection on point clouds. Our code is publicly available at \url{https://github.com/sirolf-otrebla/CompositeNet}.
Authors: Esther Julien, Krzysztof Postek, \c{S}. \.Ilker Birbil
Abstract: Two-stage robust optimization problems constitute one of the hardest optimization problem classes. One of the solution approaches to this class of problems is K-adaptability. This approach simultaneously seeks the best partitioning of the uncertainty set of scenarios into K subsets, and optimizes decisions corresponding to each of these subsets. In general case, it is solved using the K-adaptability branch-and-bound algorithm, which requires exploration of exponentially-growing solution trees. To accelerate finding high-quality solutions in such trees, we propose a machine learning-based node selection strategy. In particular, we construct a feature engineering scheme based on general two-stage robust optimization insights that allows us to train our machine learning tool on a database of resolved B&B trees, and to apply it as-is to problems of different sizes and/or types. We experimentally show that using our learned node selection strategy outperforms a vanilla, random node selection strategy when tested on problems of the same type as the training problems, also in case the K-value or the problem size differs from the training ones.
Authors: Jonas Sander, Sebastian Berndt, Ida Bruhns, Thomas Eisenbarth
Abstract: The adoption of machine learning solutions is rapidly increasing across all parts of society. As the models grow larger, both training and inference of machine learning models is increasingly outsourced, e.g. to cloud service providers. This means that potentially sensitive data is processed on untrusted platforms, which bears inherent data security and privacy risks. In this work, we investigate how to protect distributed machine learning systems, focusing on deep convolutional neural networks. The most common and best-performing mixed MPC approaches are based on HE, secret sharing, and garbled circuits. They commonly suffer from large performance overheads, big accuracy losses, and communication overheads that grow linearly in the depth of the neural network. To improve on these problems, we present Dash, a fast and distributed private convolutional neural network inference scheme secure against malicious attackers. Building on arithmetic garbling gadgets [BMR16] and fancy-garbling [BCM+19], Dash is based purely on arithmetic garbled circuits. We introduce LabelTensors that allow us to leverage the massive parallelity of modern GPUs. Combined with state-of-the-art garbling optimizations, Dash outperforms previous garbling approaches up to a factor of about 100. Furthermore, we introduce an efficient scaling operation over the residues of the Chinese remainder theorem representation to arithmetic garbled circuits, which allows us to garble larger networks and achieve much higher accuracy than previous approaches. Finally, Dash requires only a single communication round per inference step, regardless of the depth of the neural network, and a very small constant online communication volume.
Authors: Julian Coda-Forno, Kristin Witte, Akshay K. Jagadish, Marcel Binz, Zeynep Akata, Eric Schulz
Abstract: Large language models (LLMs) are transforming research on machine learning while galvanizing public debates. Understanding not only when these models work well and succeed but also why they fail and misbehave is of great societal relevance. We propose to turn the lens of psychiatry, a framework used to describe and modify maladaptive behavior, to the outputs produced by these models. We focus on twelve established LLMs and subject them to a questionnaire commonly used in psychiatry. Our results show that six of the latest LLMs respond robustly to the anxiety questionnaire, producing comparable anxiety scores to humans. Moreover, the LLMs' responses can be predictably changed by using anxiety-inducing prompts. Anxiety-induction not only influences LLMs' scores on an anxiety questionnaire but also influences their behavior in a previously-established benchmark measuring biases such as racism and ageism. Importantly, greater anxiety-inducing text leads to stronger increases in biases, suggesting that how anxiously a prompt is communicated to large language models has a strong influence on their behavior in applied settings. These results demonstrate the usefulness of methods taken from psychiatry for studying the capable algorithms to which we increasingly delegate authority and autonomy.
Authors: Felix Benning, Leif D\"oring
Abstract: Classical worst-case optimization theory neither explains the success of optimization in machine learning, nor does it help with step size selection. In this paper we demonstrate the viability and advantages of replacing the classical 'convex function' framework with a 'random function' framework. With complexity $\mathcal{O}(n^3d^3)$, where $n$ is the number of steps and $d$ the number of dimensions, Bayesian optimization with gradients has not been viable in large dimension so far. By bridging the gap between Bayesian optimization (i.e. random function optimization theory) and classical optimization we establish viability. Specifically, we use a 'stochastic Taylor approximation' to rediscover gradient descent, which is scalable in high dimension due to $\mathcal{O}(nd)$ complexity. This rediscovery yields a specific step size schedule we call Random Function Descent (RFD). The advantage of this random function framework is that RFD is scale invariant and that it provides a theoretical foundation for common step size heuristics such as gradient clipping and gradual learning rate warmup.
Authors: Marcin Przewi\k{e}\'zlikowski, Mateusz Pyla, Bartosz Zieli\'nski, Bart{\l}omiej Twardowski, Jacek Tabor, Marek \'Smieja
Abstract: Self-supervised learning (SSL) is a powerful technique for learning from unlabeled data. By learning to remain invariant to applied data augmentations, methods such as SimCLR and MoCo can reach quality on par with supervised approaches. However, this invariance may be detrimental for solving downstream tasks that depend on traits affected by augmentations used during pretraining, such as color. In this paper, we propose to foster sensitivity to such characteristics in the representation space by modifying the projector network, a common component of self-supervised architectures. Specifically, we supplement the projector with information about augmentations applied to images. For the projector to take advantage of this auxiliary conditioning when solving the SSL task, the feature extractor learns to preserve the augmentation information in its representations. Our approach, coined Conditional Augmentation-aware Self-supervised Learning (CASSLE), is directly applicable to typical joint-embedding SSL methods regardless of their objective functions. Moreover, it does not require major changes in the network architecture or prior knowledge of downstream tasks. In addition to an analysis of sensitivity towards different data augmentations, we conduct a series of experiments, which show that CASSLE improves over various SSL methods, reaching state-of-the-art performance in multiple downstream tasks.
Authors: Jo\~ao A. Leite, Olesya Razuvayevskaya, Kalina Bontcheva, Carolina Scarton
Abstract: Credibility signals represent a wide range of heuristics typically used by journalists and fact-checkers to assess the veracity of online content. Automating the extraction of credibility signals presents significant challenges due to the necessity of training high-accuracy, signal-specific extractors, coupled with the lack of sufficiently large annotated datasets. This paper introduces Pastel (Prompted weAk Supervision wiTh crEdibility signaLs), a weakly supervised approach that leverages large language models (LLMs) to extract credibility signals from web content, and subsequently combines them to predict the veracity of content without relying on human supervision. We validate our approach using four article-level misinformation detection datasets, demonstrating that Pastel outperforms zero-shot veracity detection by 38.3% and achieves 86.7% of the performance of the state-of-the-art system trained with human supervision. Moreover, in cross-domain settings where training and testing datasets originate from different domains, Pastel significantly outperforms the state-of-the-art supervised model by 63%. We further study the association between credibility signals and veracity, and perform an ablation study showing the impact of each signal on model performance. Our findings reveal that 12 out of the 19 proposed signals exhibit strong associations with veracity across all datasets, while some signals show domain-specific strengths.
Authors: Yen-Jen Wang, Bike Zhang, Jianyu Chen, Koushil Sreenath
Abstract: Large language models (LLMs) pre-trained on vast internet-scale data have showcased remarkable capabilities across diverse domains. Recently, there has been escalating interest in deploying LLMs for robotics, aiming to harness the power of foundation models in real-world settings. However, this approach faces significant challenges, particularly in grounding these models in the physical world and in generating dynamic robot motions. To address these issues, we introduce a novel paradigm in which we use few-shot prompts collected from the physical environment, enabling the LLM to autoregressively generate low-level control commands for robots without task-specific fine-tuning. Experiments across various robots and environments validate that our method can effectively prompt a robot to walk. We thus illustrate how LLMs can proficiently function as low-level feedback controllers for dynamic motion control even in high-dimensional robotic systems. The project website and source code can be found at: https://prompt2walk.github.io/ .
Authors: Deniz Bayazit, Negar Foroutan, Zeming Chen, Gail Weiss, Antoine Bosselut
Abstract: Pretrained language models (LMs) encode implicit representations of knowledge in their parameters. However, localizing these representations and disentangling them from each other remains an open problem. In this work, we investigate whether pretrained language models contain various knowledge-critical subnetworks: particular sparse computational subgraphs that can, if removed, precisely suppress specific knowledge the model has memorized. We propose a multi-objective differentiable masking scheme that can be applied to both weights and neurons to discover such subnetworks and show that we can use them to precisely remove specific knowledge from models while minimizing adverse effects on the behavior of the original model. We demonstrate our method on multiple GPT2 variants, uncovering highly sparse subnetworks (98%+ sparsity) that are critical for expressing specific collections of relational knowledge. When these subnetworks are removed, the remaining network maintains most of its initial abilities but struggles to represent the suppressed knowledge.
Authors: Huiling Zhang, Zi Xu
Abstract: Stochastic nonconvex minimax problems have attracted wide attention in machine learning, signal processing and many other fields in recent years. In this paper, we propose an accelerated first-order regularized momentum descent ascent algorithm (FORMDA) for solving stochastic nonconvex-concave minimax problems. The iteration complexity of the algorithm is proved to be $\tilde{\mathcal{O}}(\varepsilon ^{-6.5})$ to obtain an $\varepsilon$-stationary point, which achieves the best-known complexity bound for single-loop algorithms to solve the stochastic nonconvex-concave minimax problems under the stationarity of the objective function.
Authors: Biyonka Liang, Iavor Bojinov
Abstract: Experimentation is crucial for managers to rigorously quantify the value of a change and determine if it leads to a statistically significant improvement over the status quo. As companies increasingly mandate that all changes undergo experimentation before widespread release, two challenges arise: (1) minimizing the proportion of customers assigned to the inferior treatment and (2) increasing experimentation velocity by enabling data-dependent stopping. This paper addresses both challenges by introducing the Mixture Adaptive Design (MAD), a new experimental design for multi-armed bandit (MAB) algorithms that enables anytime-valid inference on the Average Treatment Effect (ATE) for \emph{any} MAB algorithm. Intuitively, MAD "mixes" any bandit algorithm with a Bernoulli design, where at each time step, the probability of assigning a unit via the Bernoulli design is determined by a user-specified deterministic sequence that can converge to zero. This sequence lets managers directly control the trade-off between regret minimization and inferential precision. Under mild conditions on the rate the sequence converges to zero, we provide a confidence sequence that is asymptotically anytime-valid and guaranteed to shrink around the true ATE. Hence, when the true ATE converges to a non-zero value, the MAD confidence sequence is guaranteed to exclude zero in finite time. Therefore, the MAD enables managers to stop experiments early while ensuring valid inference, enhancing both the efficiency and reliability of adaptive experiments. Empirically, we demonstrate that the MAD achieves finite-sample anytime-validity while accurately and precisely estimating the ATE, all without incurring significant losses in reward compared to standard bandit designs.
Authors: Shpresim Sadiku, Moritz Wagner, Sebastian Pokutta
Abstract: Sparse adversarial attacks fool deep neural networks (DNNs) through minimal pixel perturbations, often regularized by the $\ell_0$ norm. Recent efforts have replaced this norm with a structural sparsity regularizer, such as the nuclear group norm, to craft group-wise sparse adversarial attacks. The resulting perturbations are thus explainable and hold significant practical relevance, shedding light on an even greater vulnerability of DNNs. However, crafting such attacks poses an optimization challenge, as it involves computing norms for groups of pixels within a non-convex objective. We address this by presenting a two-phase algorithm that generates group-wise sparse attacks within semantically meaningful areas of an image. Initially, we optimize a quasinorm adversarial loss using the $1/2-$quasinorm proximal operator tailored for non-convex programming. Subsequently, the algorithm transitions to a projected Nesterov's accelerated gradient descent with $2-$norm regularization applied to perturbation magnitudes. Rigorous evaluations on CIFAR-10 and ImageNet datasets demonstrate a remarkable increase in group-wise sparsity, e.g., $50.9\%$ on CIFAR-10 and $38.4\%$ on ImageNet (average case, targeted attack). This performance improvement is accompanied by significantly faster computation times, improved explainability, and a $100\%$ attack success rate.
Authors: Ruihan Yang, Yejin Kim, Rose Hendrix, Aniruddha Kembhavi, Xiaolong Wang, Kiana Ehsani
Abstract: Recent advancements in robotics have enabled robots to navigate complex scenes or manipulate diverse objects independently. However, robots are still impotent in many household tasks requiring coordinated behaviors such as opening doors. The factorization of navigation and manipulation, while effective for some tasks, fails in scenarios requiring coordinated actions. To address this challenge, we introduce, HarmonicMM, an end-to-end learning method that optimizes both navigation and manipulation, showing notable improvement over existing techniques in everyday tasks. This approach is validated in simulated and real-world environments and adapts to novel unseen settings without additional tuning. Our contributions include a new benchmark for mobile manipulation and the successful deployment with only RGB visual observation in a real unseen apartment, demonstrating the potential for practical indoor robot deployment in daily life. More results are on our project site: https://rchalyang.github.io/HarmonicMM/
Authors: Henry Bae, Aghyad Deeb, Alex Fleury, Kehang Zhu
Abstract: We present ComplexityNet, a streamlined language model designed for assessing task complexity. This model predicts the likelihood of accurate output by various language models, each with different capabilities. Our initial application of ComplexityNet involves the Mostly Basic Python Problems (MBPP) dataset. We pioneered the creation of the first set of labels to define task complexity. ComplexityNet achieved a notable 79% accuracy in determining task complexity, a significant improvement over the 34% accuracy of the original, non fine-tuned model. Furthermore, ComplexityNet effectively reduces computational resource usage by 90% compared to using the highest complexity model, while maintaining a high code generation accuracy of 86.7%. This study demonstrates that fine-tuning smaller models to categorize tasks based on their complexity can lead to a more balanced trade-off between accuracy and efficiency in the use of Large Language Models. Our findings suggest a promising direction for optimizing LLM applications, especially in resource-constrained environments.
Authors: Fu-Yun Wang, Zhaoyang Huang, Xiaoyu Shi, Weikang Bian, Keqiang Sun, Guanglu Song, Yu Liu, Hongsheng Li
Abstract: This paper introduces an effective method for computation-efficient personalized style video generation without requiring access to any personalized video data. It reduces the necessary generation time of similarly sized video diffusion models from 25 seconds to around 1 second while maintaining the same level of performance. The method's effectiveness lies in its dual-level decoupling learning approach: 1) separating the learning of video style from video generation acceleration, which allows for personalized style video generation without any personalized style video data, and 2) separating the acceleration of image generation from the acceleration of video motion generation, enhancing training efficiency and mitigating the negative effects of low-quality video data.
Authors: Zihan Wang, Yunxuan Li, Yuexin Wu, Liangchen Luo, Le Hou, Hongkun Yu, Jingbo Shang
Abstract: Process supervision, using a trained verifier to evaluate the intermediate steps generated by a reasoner, has demonstrated significant improvements in multi-step problem solving. In this paper, to avoid the expensive effort of human annotation on the verifier training data, we introduce Model-induced Process Supervision (MiPS), a novel method for automating data curation. MiPS annotates an intermediate step by sampling completions of this solution through the reasoning model, and obtaining an accuracy defined as the proportion of correct completions. Inaccuracies of the reasoner would cause MiPS underestimating the accuracy of intermediate steps, therefore, we suggest and empirically show that verification focusing on high predicted scores of the verifier shall be preferred over that of low predicted scores, contrary to prior observations on human curated data. Our approach significantly improves the performance of PaLM 2 on math and coding tasks (accuracy +0.67% on GSM8K, +4.16% on MATH, +0.92% on MBPP compared with an output supervision trained verifier). Additionally, our study demonstrates that the verifier exhibits strong generalization ability across different reasoning models.
Authors: Jan Wehner, Frans Oliehoek, Luciano Cavalcante Siebert
Abstract: Learning rewards from human behaviour or feedback is a promising approach to aligning AI systems with human values but fails to consistently extract correct reward functions. Interpretability tools could enable users to understand and evaluate possible flaws in learned reward functions. We propose Counterfactual Trajectory Explanations (CTEs) to interpret reward functions in reinforcement learning by contrasting an original with a counterfactual partial trajectory and the rewards they each receive. We derive six quality criteria for CTEs and propose a novel Monte-Carlo-based algorithm for generating CTEs that optimises these quality criteria. Finally, we measure how informative the generated explanations are to a proxy-human model by training it on CTEs. CTEs are demonstrably informative for the proxy-human model, increasing the similarity between its predictions and the reward function on unseen trajectories. Further, it learns to accurately judge differences in rewards between trajectories and generalises to out-of-distribution examples. Although CTEs do not lead to a perfect understanding of the reward, our method, and more generally the adaptation of XAI methods, are presented as a fruitful approach for interpreting learned reward functions.
Authors: Jacob Russin, Ellie Pavlick, Michael J. Frank
Abstract: Human learning embodies a striking duality: sometimes, we appear capable of following logical, compositional rules and benefit from structured curricula (e.g., in formal education), while other times, we rely on an incremental approach or trial-and-error, learning better from curricula that are unstructured or randomly interleaved. Influential psychological theories explain this seemingly disparate behavioral evidence by positing two qualitatively different learning systems -- one for rapid, rule-based inferences and another for slow, incremental adaptation. It remains unclear how to reconcile such theories with neural networks, which learn via incremental weight updates and are thus a natural model for the latter type of learning, but are not obviously compatible with the former. However, recent evidence suggests that both metalearning neural networks and large language models are capable of "in-context learning" (ICL) -- the ability to flexibly grasp the structure of a new task from a few examples given at inference time. Here, we show that networks capable of ICL can reproduce human-like learning and compositional behavior on rule-governed tasks, while at the same time replicating human behavioral phenomena in tasks lacking rule-like structure via their usual in-weight learning (IWL). Our work shows how emergent ICL can equip neural networks with fundamentally different learning properties than those traditionally attributed to them, and that these can coexist with the properties of their native IWL, thus offering a novel perspective on dual-process theories and human cognitive flexibility.
Authors: Sirui Hong, Yizhang Lin, Bang Liu, Bangbang Liu, Binhao Wu, Ceyao Zhang, Chenxing Wei, Danyang Li, Jiaqi Chen, Jiayi Zhang, Jinlin Wang, Li Zhang, Lingyao Zhang, Min Yang, Mingchen Zhuge, Taicheng Guo, Tuo Zhou, Wei Tao, Xiangru Tang, Xiangtao Lu, Xiawu Zheng, Xinbing Liang, Yaying Fei, Yuheng Cheng, Zhibin Gou, Zongze Xu, Chenglin Wu
Abstract: Large Language Model (LLM)-based agents have shown effectiveness across many applications. However, their use in data science scenarios requiring solving long-term interconnected tasks, dynamic data adjustments and domain expertise remains challenging. Previous approaches primarily focus on individual tasks, making it difficult to assess the complete data science workflow. Moreover, they struggle to handle real-time changes in intermediate data and fail to adapt dynamically to evolving task dependencies inherent to data science problems. In this paper, we present Data Interpreter, an LLM-based agent designed to automatically solve various data science problems end-to-end. Our Data Interpreter incorporates two key modules: 1) Hierarchical Graph Modeling, which breaks down complex problems into manageable subproblems, enabling dynamic node generation and graph optimization; and 2) Programmable Node Generation, a technique that refines and verifies each subproblem to iteratively improve code generation results and robustness. Extensive experiments consistently demonstrate the superiority of Data Interpreter. On InfiAgent-DABench, it achieves a 25% performance boost, raising accuracy from 75.9% to 94.9%. For machine learning and open-ended tasks, it improves performance from 88% to 95%, and from 60% to 97%, respectively. Moreover, on the MATH dataset, Data Interpreter achieves remarkable performance with a 26% improvement compared to state-of-the-art baselines. The code is available at https://github.com/geekan/MetaGPT.
Authors: Austin E. Y. T. Lefebvre (Calico Life Sciences LLC), Gabriel Sturm (Calico Life Sciences LLC, Department of Biochemistry and Biophysics, University of California San Francisco), Ting-Yu Lin (Calico Life Sciences LLC), Emily Stoops (Calico Life Sciences LLC), Magdalena Preciado Lopez (Calico Life Sciences LLC), Benjamin Kaufmann-Malaga (Calico Life Sciences LLC), Kayley Hake (Calico Life Sciences LLC)
Abstract: The analysis of dynamic organelles remains a formidable challenge, though key to understanding biological processes. We introduce Nellie, an automated and unbiased user-friendly pipeline for segmentation, tracking, and feature extraction of diverse intracellular structures. Nellie adapts to image metadata, eliminating user input. Nellie's preprocessing pipeline enhances structural contrast on multiple intracellular scales allowing for robust hierarchical segmentation of sub-organellar regions. Internal motion capture markers are generated and tracked via a radius-adaptive pattern matching scheme, and used as guides for sub-voxel flow interpolation. Nellie extracts a plethora of features at multiple hierarchical levels for deep and customizable analysis. Nellie features a point-and-click Napari-based GUI that allows for code-free operation and visualization, while its modular open-source codebase invites extension by experienced users. We demonstrate Nellie's wide variety of use cases with three examples: unmixing multiple organelles from a single channel using feature-based classification, training an unsupervised graph autoencoder on mitochondrial multi-mesh graphs to quantify latent space embedding changes following ionomycin treatment, and performing in-depth characterization and comparison of endoplasmic reticulum networks across different cell types and temporal frames.
Authors: Luke Rowe, Roger Girgis, Anthony Gosselin, Bruno Carrez, Florian Golemo, Felix Heide, Liam Paull, Christopher Pal
Abstract: Evaluating autonomous vehicle stacks (AVs) in simulation typically involves replaying driving logs from real-world recorded traffic. However, agents replayed from offline data are not reactive and hard to intuitively control. Existing approaches address these challenges by proposing methods that rely on heuristics or generative models of real-world data but these approaches either lack realism or necessitate costly iterative sampling procedures to control the generated behaviours. In this work, we take an alternative approach and propose CtRL-Sim, a method that leverages return-conditioned offline reinforcement learning (RL) to efficiently generate reactive and controllable traffic agents. Specifically, we process real-world driving data through a physics-enhanced Nocturne simulator to generate a diverse offline RL dataset, annotated with various rewards. With this dataset, we train a return-conditioned multi-agent behaviour model that allows for fine-grained manipulation of agent behaviours by modifying the desired returns for the various reward components. This capability enables the generation of a wide range of driving behaviours beyond the scope of the initial dataset, including adversarial behaviours. We show that CtRL-Sim can generate realistic safety-critical scenarios while providing fine-grained control over agent behaviours.
Authors: Yuling Jiao, Lican Kang, Huazhen Lin, Jin Liu, Heng Zuo
Abstract: This paper aims to conduct a comprehensive theoretical analysis of current diffusion models. We introduce a novel generative learning methodology utilizing the Schr{\"o}dinger bridge diffusion model in latent space as the framework for theoretical exploration in this domain. Our approach commences with the pre-training of an encoder-decoder architecture using data originating from a distribution that may diverge from the target distribution, thus facilitating the accommodation of a large sample size through the utilization of pre-existing large-scale models. Subsequently, we develop a diffusion model within the latent space utilizing the Schr{\"o}dinger bridge framework. Our theoretical analysis encompasses the establishment of end-to-end error analysis for learning distributions via the latent Schr{\"o}dinger bridge diffusion model. Specifically, we control the second-order Wasserstein distance between the generated distribution and the target distribution. Furthermore, our obtained convergence rates effectively mitigate the curse of dimensionality, offering robust theoretical support for prevailing diffusion models.
Authors: Xiaofei Wang, Yunfeng Zhao, Chao Qiu, Qinghua Hu, Victor C. M. Leung
Abstract: Amidst the robust impetus from artificial intelligence (AI) and big data, edge intelligence (EI) has emerged as a nascent computing paradigm, synthesizing AI with edge computing (EC) to become an exemplary solution for unleashing the full potential of AI services. Nonetheless, challenges in communication costs, resource allocation, privacy, and security continue to constrain its proficiency in supporting services with diverse requirements. In response to these issues, this paper introduces socialized learning (SL) as a promising solution, further propelling the advancement of EI. SL is a learning paradigm predicated on social principles and behaviors, aimed at amplifying the collaborative capacity and collective intelligence of agents within the EI system. SL not only enhances the system's adaptability but also optimizes communication, and networking processes, essential for distributed intelligence across diverse devices and platforms. Therefore, a combination of SL and EI may greatly facilitate the development of collaborative intelligence in the future network. This paper presents the findings of a literature review on the integration of EI and SL, summarizing the latest achievements in existing research on EI and SL. Subsequently, we delve comprehensively into the limitations of EI and how it could benefit from SL. Special emphasis is placed on the communication challenges and networking strategies and other aspects within these systems, underlining the role of optimized network solutions in improving system efficiency. Based on these discussions, we elaborate in detail on three integrated components: socialized architecture, socialized training, and socialized inference, analyzing their strengths and weaknesses. Finally, we identify some possible future applications of combining SL and EI, discuss open problems and suggest some future research.
Authors: Guanglin Zhou, Zhongyi Han, Shiming Chen, Biwei Huang, Liming Zhu, Salman Khan, Xin Gao, Lina Yao
Abstract: Recent studies indicate that large multimodal models (LMMs) potentially act as general-purpose assistants and are highly robust against different distributions. Despite this, domain-specific adaptation is still necessary particularly in specialized areas like healthcare. Due to the impracticality of fine-tuning LMMs given their vast parameter space, this work investigates in-context learning (ICL) as an effective alternative for enhancing LMMs' adaptability. Our study addresses this by evaluating an unsupervised ICL method which selects in-context examples through a nearest example search based on feature similarity. We uncover that its effectiveness is limited by the deficiencies of pre-trained vision encoders under distribution shift scenarios. To address these challenges, we propose InvariantSelectPR, a novel method leveraging Class-conditioned Contrastive Invariance (CCI) for more robust demonstration selection. Specifically, CCI enhances pre-trained vision encoders by improving their discriminative capabilities across different classes and ensuring invariance to domain-specific variations. This enhancement allows the encoders to effectively identify and retrieve the most informative examples, which are then used to guide LMMs in adapting to new query samples under varying distributions. Our experiments show that InvariantSelectPR substantially improves the adaptability of LMMs, achieving significant performance gains on benchmark datasets, with a 34.2%$\uparrow$ accuracy increase in 7-shot on Camelyon17 and 16.9%$\uparrow$ increase in 7-shot on HAM10000 compared to the baseline zero-shot performance.
Authors: Shuofei Qiao, Runnan Fang, Ningyu Zhang, Yuqi Zhu, Xiang Chen, Shumin Deng, Yong Jiang, Pengjun Xie, Fei Huang, Huajun Chen
Abstract: Recent endeavors towards directly using large language models (LLMs) as agent models to execute interactive planning tasks have shown commendable results. Despite their achievements, however, they still struggle with brainless trial-and-error in global planning and generating hallucinatory actions in local planning due to their poor understanding of the ``real'' physical world. Imitating humans' mental world knowledge model which provides global prior knowledge before the task and maintains local dynamic knowledge during the task, in this paper, we introduce parametric World Knowledge Model (WKM) to facilitate agent planning. Concretely, we steer the agent model to self-synthesize knowledge from both expert and sampled trajectories. Then we develop WKM, providing prior task knowledge to guide the global planning and dynamic state knowledge to assist the local planning. Experimental results on three complex real-world simulated datasets with three state-of-the-art open-source LLMs, Mistral-7B, Gemma-7B, and Llama-3-8B, demonstrate that our method can achieve superior performance compared to various strong baselines. Besides, we analyze to illustrate that our WKM can effectively alleviate the blind trial-and-error and hallucinatory action issues, providing strong support for the agent's understanding of the world. Other interesting findings include: 1) our instance-level task knowledge can generalize better to unseen tasks, 2) weak WKM can guide strong agent model planning, and 3) unified WKM training has promising potential for further development. The code is available at https://github.com/zjunlp/WKM.
Authors: Zhiwei Wang, Yunji Wang, Zhongwang Zhang, Zhangchen Zhou, Hui Jin, Tianyang Hu, Jiacheng Sun, Zhenguo Li, Yaoyu Zhang, Zhi-Qin John Xu
Abstract: Large language models have consistently struggled with complex reasoning tasks, such as mathematical problem-solving. Investigating the internal reasoning mechanisms of these models can help us design better model architectures and training strategies, ultimately enhancing their reasoning capability. In this study, we constructed a symbolic dataset to investigate the mechanisms by which Transformer models employ vertical thinking strategy based on their inherent structure and horizontal thinking strategy based on Chain of Thought to achieve multi-step reasoning. We introduced the concept of buffer mechanism: the model stores various information in distinct buffers and selectively extracts them through the query-key matrix. We proposed a random matrix-based algorithm to enhance the model's reasoning ability, resulting in a 75% reduction in the training time required for the GPT-2 model to achieve generalization capability on the PrOntoQA dataset. These findings provide new insights into understanding the mechanisms of large language models.
Authors: Harsh Chaudhari, Giorgio Severi, John Abascal, Matthew Jagielski, Christopher A. Choquette-Choo, Milad Nasr, Cristina Nita-Rotaru, Alina Oprea
Abstract: Retrieval Augmented Generation (RAG) expands the capabilities of modern large language models (LLMs), by anchoring, adapting, and personalizing their responses to the most relevant knowledge sources. It is particularly useful in chatbot applications, allowing developers to customize LLM output without expensive retraining. Despite their significant utility in various applications, RAG systems present new security risks. In this work, we propose new attack vectors that allow an adversary to inject a single malicious document into a RAG system's knowledge base, and mount a backdoor poisoning attack. We design Phantom, a general two-stage optimization framework against RAG systems, that crafts a malicious poisoned document leading to an integrity violation in the model's output. First, the document is constructed to be retrieved only when a specific trigger sequence of tokens appears in the victim's queries. Second, the document is further optimized with crafted adversarial text that induces various adversarial objectives on the LLM output, including refusal to answer, reputation damage, privacy violations, and harmful behaviors. We demonstrate our attacks on multiple LLM architectures, including Gemma, Vicuna, and Llama, and show that they transfer to GPT-3.5 Turbo and GPT-4. Finally, we successfully conducted a Phantom attack on NVIDIA's black-box production RAG system, "Chat with RTX".
Authors: Yijiong Yu, Huiqiang Jiang, Xufang Luo, Qianhui Wu, Chin-Yew Lin, Dongsheng Li, Yuqing Yang, Yongfeng Huang, Lili Qiu
Abstract: Large Language Models (LLMs) are increasingly applied in various real-world scenarios due to their excellent generalization capabilities and robust generative abilities. However, they exhibit position bias, also known as "lost in the middle", a phenomenon that is especially pronounced in long-context scenarios, which indicates the placement of the key information in different positions of a prompt can significantly affect accuracy. This paper first explores the micro-level manifestations of position bias, concluding that attention weights are a micro-level expression of position bias. It further identifies that, in addition to position embeddings, causal attention mask also contributes to position bias by creating position-specific hidden states. Based on these insights, we propose a method to mitigate position bias by scaling this positional hidden states. Experiments on the NaturalQuestions Multi-document QA, KV retrieval, LongBench and timeline reorder tasks, using various models including RoPE models, context windowextended models, and Alibi models, demonstrate the effectiveness and generalizability of our approach. Our method can improve performance by up to 15.2% by modifying just one dimension of hidden states. Our code is available at https://aka.ms/PositionalHidden.
Authors: Heejun Lee, Geon Park, Youngwan Lee, Jaduk Suh, Jina Kim, Wonyoung Jeong, Bumsik Kim, Hyemin Lee, Myeongjae Jeon, Sung Ju Hwang
Abstract: In modern large language models (LLMs), increasing the context length is crucial for improving comprehension and coherence in long-context, multi-modal, and retrieval-augmented language generation. While many recent transformer models attempt to extend their context length over a million tokens, they remain impractical due to the quadratic time and space complexities. Although recent works on linear and sparse attention mechanisms can achieve this goal, their real-world applicability is often limited by the need to re-train from scratch and significantly worse performance. In response, we propose a novel approach, Hierarchically Pruned Attention (HiP), which reduces the time complexity of the attention mechanism to $O(T \log T)$ and the space complexity to $O(T)$, where $T$ is the sequence length. We notice a pattern in the attention scores of pretrained LLMs where tokens close together tend to have similar scores, which we call ``attention locality''. Based on this observation, we utilize a novel tree-search-like algorithm that estimates the top-$k$ key tokens for a given query on the fly, which is mathematically guaranteed to have better performance than random attention pruning. In addition to improving the time complexity of the attention mechanism, we further optimize GPU memory usage by implementing KV cache offloading, which stores only $O(\log T)$ tokens on the GPU while maintaining similar decoding throughput. Experiments on benchmarks show that HiP, with its training-free nature, significantly reduces both prefill and decoding latencies, as well as memory usage, while maintaining high-quality generation with minimal degradation. HiP enables pretrained LLMs to scale up to millions of tokens on commodity GPUs, potentially unlocking long-context LLM applications previously deemed infeasible.
Authors: Mehar Khurana, Neehar Peri, James Hays, Deva Ramanan
Abstract: State-of-the-art 3D object detectors are often trained on massive labeled datasets. However, annotating 3D bounding boxes remains prohibitively expensive and time-consuming, particularly for LiDAR. Instead, recent works demonstrate that self-supervised pre-training with unlabeled data can improve detection accuracy with limited labels. Contemporary methods adapt best-practices for self-supervised learning from the image domain to point clouds (such as contrastive learning). However, publicly available 3D datasets are considerably smaller and less diverse than those used for image-based self-supervised learning, limiting their effectiveness. We do note, however, that such 3D data is naturally collected in a multimodal fashion, often paired with images. Rather than pre-training with only self-supervised objectives, we argue that it is better to bootstrap point cloud representations using image-based foundation models trained on internet-scale data. Specifically, we propose a shelf-supervised approach (e.g. supervised with off-the-shelf image foundation models) for generating zero-shot 3D bounding boxes from paired RGB and LiDAR data. Pre-training 3D detectors with such pseudo-labels yields significantly better semi-supervised detection accuracy than prior self-supervised pretext tasks. Importantly, we show that image-based shelf-supervision is helpful for training LiDAR-only, RGB-only and multi-modal (RGB + LiDAR) detectors. We demonstrate the effectiveness of our approach on nuScenes and WOD, significantly improving over prior work in limited data settings. Our code is available at https://github.com/meharkhurana03/cm3d
Authors: Amit Das, Zheng Zhang, Najib Hasan, Souvika Sarkar, Fatemeh Jamshidi, Tathagata Bhattacharya, Mostafa Rahgouy, Nilanjana Raychawdhary, Dongji Feng, Vinija Jain, Aman Chadha, Mary Sandage, Lauramarie Pope, Gerry Dozier, Cheryl Seals
Abstract: Data annotation, the practice of assigning descriptive labels to raw data, is pivotal in optimizing the performance of machine learning models. However, it is a resource-intensive process susceptible to biases introduced by annotators. The emergence of sophisticated Large Language Models (LLMs) presents a unique opportunity to modernize and streamline this complex procedure. While existing research extensively evaluates the efficacy of LLMs, as annotators, this paper delves into the biases present in LLMs when annotating hate speech data. Our research contributes to understanding biases in four key categories: gender, race, religion, and disability with four LLMs: GPT-3.5, GPT-4o, Llama-3.1 and Gemma-2. Specifically targeting highly vulnerable groups within these categories, we analyze annotator biases. Furthermore, we conduct a comprehensive examination of potential factors contributing to these biases by scrutinizing the annotated data. We introduce our custom hate speech detection dataset, HateBiasNet, to conduct this research. Additionally, we perform the same experiments on the ETHOS (Mollas et al. 2022) dataset also for comparative analysis. This paper serves as a crucial resource, guiding researchers and practitioners in harnessing the potential of LLMs for data annotation, thereby fostering advancements in this critical field.
Authors: Eldar Kurtic, Amir Moeini, Dan Alistarh
Abstract: We introduce Mathador-LM, a new benchmark for evaluating the mathematical reasoning on large language models (LLMs), combining ruleset interpretation, planning, and problem-solving. This benchmark is inspired by the Mathador game, where the objective is to reach a target number using basic arithmetic operations on a given set of base numbers, following a simple set of rules. We show that, across leading LLMs, we obtain stable average performance while generating benchmark instances \emph{dynamically}, following a target difficulty level. Thus, our benchmark alleviates concerns about test-set leakage into training data, an issue that often undermines popular benchmarks. Additionally, we conduct a comprehensive evaluation of both open and closed-source state-of-the-art LLMs on Mathador-LM. Our findings reveal that contemporary models struggle with Mathador-LM, scoring significantly lower than average 3rd graders. This stands in stark contrast to their strong performance on popular mathematical reasoning benchmarks. The implementation of Mathador-LM benchmark is available at \href{https://github.com/IST-DASLab/Mathador-LM}{github.com/IST-DASLab/Mathador-LM}.
Authors: Guido Di Federico, Louis J. Durlofsky
Abstract: Geological parameterization entails the representation of a geomodel using a small set of latent variables and a mapping from these variables to grid-block properties such as porosity and permeability. Parameterization is useful for data assimilation (history matching), as it maintains geological realism while reducing the number of variables to be determined. Diffusion models are a new class of generative deep-learning procedures that have been shown to outperform previous methods, such as generative adversarial networks, for image generation tasks. Diffusion models are trained to "denoise", which enables them to generate new geological realizations from input fields characterized by random noise. Latent diffusion models, which are the specific variant considered in this study, provide dimension reduction through use of a low-dimensional latent variable. The model developed in this work includes a variational autoencoder for dimension reduction and a U-net for the denoising process. Our application involves conditional 2D three-facies (channel-levee-mud) systems. The latent diffusion model is shown to provide realizations that are visually consistent with samples from geomodeling software. Quantitative metrics involving spatial and flow-response statistics are evaluated, and general agreement between the diffusion-generated models and reference realizations is observed. Stability tests are performed to assess the smoothness of the parameterization method. The latent diffusion model is then used for ensemble-based data assimilation. Two synthetic "true" models are considered. Significant uncertainty reduction, posterior P$_{10}$-P$_{90}$ forecasts that generally bracket observed data, and consistent posterior geomodels, are achieved in both cases.
Authors: Dian Wang, Stephen Hart, David Surovik, Tarik Kelestemur, Haojie Huang, Haibo Zhao, Mark Yeatman, Jiuguang Wang, Robin Walters, Robert Platt
Abstract: Recent work has shown diffusion models are an effective approach to learning the multimodal distributions arising from demonstration data in behavior cloning. However, a drawback of this approach is the need to learn a denoising function, which is significantly more complex than learning an explicit policy. In this work, we propose Equivariant Diffusion Policy, a novel diffusion policy learning method that leverages domain symmetries to obtain better sample efficiency and generalization in the denoising function. We theoretically analyze the $\mathrm{SO}(2)$ symmetry of full 6-DoF control and characterize when a diffusion model is $\mathrm{SO}(2)$-equivariant. We furthermore evaluate the method empirically on a set of 12 simulation tasks in MimicGen, and show that it obtains a success rate that is, on average, 21.9% higher than the baseline Diffusion Policy. We also evaluate the method on a real-world system to show that effective policies can be learned with relatively few training samples, whereas the baseline Diffusion Policy cannot.
Authors: Alejandro Rodriguez-Garcia, Jie Mei, Srikanth Ramaswamy
Abstract: Recent progress in artificial intelligence (AI) has been driven by insights from neuroscience, particularly with the development of artificial neural networks (ANNs). This has significantly enhanced the replication of complex cognitive tasks such as vision and natural language processing. Despite these advances, ANNs struggle with continual learning, adaptable knowledge transfer, robustness, and resource efficiency - capabilities that biological systems handle seamlessly. Specifically, ANNs often overlook the functional and morphological diversity of the brain, hindering their computational capabilities. Furthermore, incorporating cell-type specific neuromodulatory effects into ANNs with neuronal heterogeneity could enable learning at two spatial scales: spiking behavior at the neuronal level, and synaptic plasticity at the circuit level, thereby potentially enhancing their learning abilities. In this article, we summarize recent bio-inspired models, learning rules and architectures and propose a biologically-informed framework for enhancing ANNs. Our proposed dual-framework approach highlights the potential of spiking neural networks (SNNs) for emulating diverse spiking behaviors and dendritic compartments to simulate morphological and functional diversity of neuronal computations. Finally, we outline how the proposed approach integrates brain-inspired compartmental models and task-driven SNNs, balances bioinspiration and complexity, and provides scalable solutions for pressing AI challenges, such as continual learning, adaptability, robustness, and resource-efficiency.
Authors: Matteo Belenchia, Flavio Corradini, Michela Quadrini, Michele Loreti
Abstract: Graph neural networks form a class of deep learning architectures specifically designed to work with graph-structured data. As such, they share the inherent limitations and problems of deep learning, especially regarding the issues of explainability and trustworthiness. We propose $\mu\mathcal{G}$, an original domain-specific language for the specification of graph neural networks that aims to overcome these issues. The language's syntax is introduced, and its meaning is rigorously defined by a denotational semantics. An equivalent characterization in the form of an operational semantics is also provided and, together with a type system, is used to prove the type soundness of $\mu\mathcal{G}$. We show how $\mu\mathcal{G}$ programs can be represented in a more user-friendly graphical visualization, and provide examples of its generality by showing how it can be used to define some of the most popular graph neural network models, or to develop any custom graph processing application.
Authors: Ishita Kumar, Snigdha Viswanathan, Sushrita Yerra, Alireza Salemi, Ryan A. Rossi, Franck Dernoncourt, Hanieh Deilamsalehy, Xiang Chen, Ruiyi Zhang, Shubham Agarwal, Nedim Lipka, Chien Van Nguyen, Thien Huu Nguyen, Hamed Zamani
Abstract: Long-text generation is seemingly ubiquitous in real-world applications of large language models such as generating an email or writing a review. Despite the fundamental importance and prevalence of long-text generation in many practical applications, existing work on personalized generation has focused on the generation of very short text. To overcome these limitations, we study the problem of personalized long-text generation, that is, generating long-text that is personalized for a specific user while being practically useful for the vast majority of real-world applications that naturally require the generation of longer text. In this work, we demonstrate the importance of user-specific personalization for long-text generation tasks and develop the Long-text Language Model Personalization (LongLaMP) Benchmark. LongLaMP provides a comprehensive and diverse evaluation framework for personalized long-text generation. Extensive experiments on LongLaMP for zero-shot and fine-tuned language tasks demonstrate the effectiveness of the proposed benchmark and its utility for developing and evaluating techniques for personalized long-text generation across a wide variety of long-text generation tasks. The results highlight the importance of personalization across a wide variety of long-text generation tasks. Finally, we release the benchmark for others to use for this important problem.
Authors: Ye Jiang, Yimin Wang
Abstract: Large visual-language models (LVLMs) exhibit exceptional performance in visual-language reasoning across diverse cross-modal benchmarks. Despite these advances, recent research indicates that Large Language Models (LLMs), like GPT-3.5-turbo, underachieve compared to well-trained smaller models, such as BERT, in Fake News Detection (FND), prompting inquiries into LVLMs' efficacy in FND tasks. Although performance could improve through fine-tuning LVLMs, the substantial parameters and requisite pre-trained weights render it a resource-heavy endeavor for FND applications. This paper initially assesses the FND capabilities of two notable LVLMs, CogVLM and GPT4V, in comparison to a smaller yet adeptly trained CLIP model in a zero-shot context. The findings demonstrate that LVLMs can attain performance competitive with that of the smaller model. Next, we integrate standard in-context learning (ICL) with LVLMs, noting improvements in FND performance, though limited in scope and consistency. To address this, we introduce the \textbf{I}n-context \textbf{M}ultimodal \textbf{F}ake \textbf{N}ews \textbf{D}etection (IMFND) framework, enriching in-context examples and test inputs with predictions and corresponding probabilities from a well-trained smaller model. This strategic integration directs the LVLMs' focus towards news segments associated with higher probabilities, thereby improving their analytical accuracy. The experimental results suggest that the IMFND framework significantly boosts the FND efficiency of LVLMs, achieving enhanced accuracy over the standard ICL approach across three publicly available FND datasets.
Authors: Sina Gholamian, Gianfranco Romani, Bartosz Rudnikowicz, Stavroula Skylaki
Abstract: Product classification is a crucial task in international trade, as compliance regulations are verified and taxes and duties are applied based on product categories. Manual classification of products is time-consuming and error-prone, and the sheer volume of products imported and exported renders the manual process infeasible. Consequently, e-commerce platforms and enterprises involved in international trade have turned to automatic product classification using machine learning. However, current approaches do not consider the real-world challenges associated with product classification, such as very abbreviated and incomplete product descriptions. In addition, recent advancements in generative Large Language Models (LLMs) and their reasoning capabilities are mainly untapped in product classification and e-commerce. In this research, we explore the real-life challenges of industrial classification and we propose data perturbations that allow for realistic data simulation. Furthermore, we employ LLM-based product classification to improve the robustness of the prediction in presence of incomplete data. Our research shows that LLMs with in-context learning outperform the supervised approaches in the clean-data scenario. Additionally, we illustrate that LLMs are significantly more robust than the supervised approaches when data attacks are present.
Authors: Yuting Liu, Jinghao Zhang, Yizhou Dang, Yuliang Liang, Qiang Liu, Guibing Guo, Jianzhe Zhao, Xingwei Wang
Abstract: Involving collaborative information in Large Language Models (LLMs) is a promising technique for adapting LLMs for recommendation. Existing methods achieve this by concatenating collaborative features with text tokens into a unified sequence input and then fine-tuning to align these features with LLM's input space. Although effective, in this work, we identify two limitations when adapting LLMs to recommendation tasks, which hinder the integration of general knowledge and collaborative information, resulting in sub-optimal recommendation performance. (1) Fine-tuning LLM with recommendation data can undermine its inherent world knowledge and fundamental competencies, which are crucial for interpreting and inferring recommendation text. (2) Incorporating collaborative features into textual prompts disrupts the semantics of the original prompts, preventing LLM from generating appropriate outputs. In this paper, we propose a new paradigm, CoRA (an acronym for Collaborative LoRA), with a collaborative weights generator. Rather than input space alignment, this method aligns collaborative information with LLM's parameter space, representing them as incremental weights to update LLM's output. This way, LLM perceives collaborative information without altering its general knowledge and text inference capabilities. Specifically, we employ a collaborative filtering model to extract user and item embeddings, converting them into collaborative weights with low-rank properties through the collaborative weights generator. We then merge the collaborative weights into LLM's weights, enabling LLM to perceive the collaborative signals and generate personalized recommendations without fine-tuning or extra collaborative tokens in prompts. Extensive experiments confirm that CoRA effectively integrates collaborative information into LLM, enhancing recommendation performance.
Authors: Maria Tsfasman, Bernd Dudzik, Kristian Fenech, Andras Lorincz, Catholijn M. Jonker, Catharine Oertel
Abstract: Conversational memory is the process by which humans encode, retain and retrieve verbal, non-verbal and contextual information from a conversation. Since human memory is selective, differing recollections of the same events can lead to misunderstandings and misalignments within a group. Yet, conversational facilitation systems, aimed at advancing the quality of group interactions, usually focus on tracking users' states within an individual session, ignoring what remains in each participant's memory after the interaction. Understanding conversational memory can be used as a source of information on the long-term development of social connections within a group. This paper introduces the MeMo corpus, the first conversational dataset annotated with participants' memory retention reports, aimed at facilitating computational modelling of human conversational memory. The MeMo corpus includes 31 hours of small-group discussions on Covid-19, repeated 3 times over the term of 2 weeks. It integrates validated behavioural and perceptual measures, audio, video, and multimodal annotations, offering a valuable resource for studying and modelling conversational memory and group dynamics. By introducing the MeMo corpus, analysing its validity, and demonstrating its usefulness for future research, this paper aims to pave the way for future research in conversational memory modelling for intelligent system development.
Authors: Jia Wan, Sean R. Sinclair, Devavrat Shah, Martin J. Wainwright
Abstract: We study a class of structured Markov Decision Processes (MDPs) known as Exo-MDPs. They are characterized by a partition of the state space into two components: the exogenous states evolve stochastically in a manner not affected by the agent's actions, whereas the endogenous states can be affected by actions, and evolve according to deterministic dynamics involving both the endogenous and exogenous states. Exo-MDPs provide a natural model for various applications, including inventory control, portfolio management, power systems, and ride-sharing, among others. While seemingly restrictive on the surface, our first result establishes that any discrete MDP can be represented as an Exo-MDP. The underlying argument reveals how transition and reward dynamics can be written as linear functions of the exogenous state distribution, showing how Exo-MDPs are instances of linear mixture MDPs, thereby showing a representational equivalence between discrete MDPs, Exo-MDPs, and linear mixture MDPs. The connection between Exo-MDPs and linear mixture MDPs leads to algorithms that are near sample-optimal, with regret guarantees scaling with the (effective) size of the exogenous state space $d$, independent of the sizes of the endogenous state and action spaces, even when the exogenous state is {\em unobserved}. When the exogenous state is unobserved, we establish a regret upper bound of $O(H^{3/2}d\sqrt{K})$ with $K$ trajectories of horizon $H$ and unobserved exogenous state of dimension $d$. We also establish a matching regret lower bound of $\Omega(H^{3/2}d\sqrt{K})$ for non-stationary Exo-MDPs and a lower bound of $\Omega(Hd\sqrt{K})$ for stationary Exo-MDPs. We complement our theoretical findings with an experimental study on inventory control problems.
Authors: Zeyue Zhang, Xiaoling Lu, Feng Zhou
Abstract: The Bayesian two-step change point detection method is popular for the Hawkes process due to its simplicity and intuitiveness. However, the non-conjugacy between the point process likelihood and the prior requires most existing Bayesian two-step change point detection methods to rely on non-conjugate inference methods. These methods lack analytical expressions, leading to low computational efficiency and impeding timely change point detection. To address this issue, this work employs data augmentation to propose a conjugate Bayesian two-step change point detection method for the Hawkes process, which proves to be more accurate and efficient. Extensive experiments on both synthetic and real data demonstrate the superior effectiveness and efficiency of our method compared to baseline methods. Additionally, we conduct ablation studies to explore the robustness of our method concerning various hyperparameters. Our code is publicly available at https://github.com/Aurora2050/CoBay-CPD.
Authors: Sathya Krishnan Suresh, Wu Mengjun, Tushar Pranav, Eng Siong Chng
Abstract: The scarcity of domain-specific dialogue datasets limits the development of dialogue systems across applications. Existing research is constrained by general or niche datasets that lack sufficient scale for training dialogue systems. To address this gap, we introduce DiaSynth - a synthetic dialogue generation framework capable of generating high-quality, contextually rich dialogues across a wide range of domains. Unlike existing frameworks, DiaSynth uses Large Language Models (LLMs) and Chain of Thought (CoT) reasoning to generate dynamic, domain-specific dialogues with simulated personas and diverse conversational features. We perform our experiments by generating synthetic data using different LLMs and few-shot examples from DialogSum and SAMSum. The pretrained language models fine-tuned on the synthetic data outperform the base models by 16.47% on dialogue summarization, while the comparison between models fine-tuned on in-domain data and synthetic data shows that the synthetic data is able to capture 90.48% of the performance distribution of the in-domain data on dialogue summarization. The quality of the data generated also increases as we increase the size of LLM from 3B to 8B. These results validate DiaSynth's potential as a robust alternative to traditional data collection methods. We open source the code and data generated for future research.
Authors: Beepul Bharti, Paul Yi, Jeremias Sulam
Abstract: As complex machine learning models continue to find applications in high-stakes decision-making scenarios, it is crucial that we can explain and understand their predictions. Post-hoc explanation methods provide useful insights by identifying important features in an input $\mathbf{x}$ with respect to the model output $f(\mathbf{x})$. In this work, we formalize and study two precise notions of feature importance for general machine learning models: sufficiency and necessity. We demonstrate how these two types of explanations, albeit intuitive and simple, can fall short in providing a complete picture of which features a model finds important. To this end, we propose a unified notion of importance that circumvents these limitations by exploring a continuum along a necessity-sufficiency axis. Our unified notion, we show, has strong ties to other popular definitions of feature importance, like those based on conditional independence and game-theoretic quantities like Shapley values. Crucially, we demonstrate how a unified perspective allows us to detect important features that could be missed by either of the previous approaches alone.
Authors: Chengpu Wang
Abstract: Statistical Taylor expansion replaces the input precise variables in a conventional Taylor expansion with random variables each with known distribution, to calculate the result mean and deviation. It is based on the uncorrelated uncertainty assumption: Each input variable is measured independently with fine enough statistical precision, so that their uncertainties are independent of each other. Statistical Taylor expansion reviews that the intermediate analytic expressions can no longer be regarded as independent of each other, and the result of analytic expression should be path independent. This conclusion differs fundamentally from the conventional common approach in applied mathematics to find the best execution path for a result. This paper also presents an implementation of statistical Taylor expansion called variance arithmetic, and the tests on variance arithmetic.
Authors: Bolun "Namir" Xia, Mohammed J. Zaki, Aparna Gupta
Abstract: The advent of large language models (LLMs) has initiated much research into their various financial applications. However, in applying LLMs on long documents, semantic relations are not explicitly incorporated, and a full or arbitrarily sparse attention operation is employed. In recent years, progress has been made in Abstract Meaning Representation (AMR), which is a graph-based representation of text to preserve its semantic relations. Since AMR can represent semantic relationships at a deeper level, it can be beneficially utilized by graph neural networks (GNNs) for constructing effective document-level graph representations built upon LLM embeddings to predict target metrics in the financial domain. We propose FLAG: Financial Long document classification via AMR-based GNN, an AMR graph based framework to generate document-level embeddings for long financial document classification. We construct document-level graphs from sentence-level AMR graphs, endow them with specialized LLM word embeddings in the financial domain, apply a deep learning mechanism that utilizes a GNN, and examine the efficacy of our AMR-based approach in predicting labeled target data from long financial documents. Extensive experiments are conducted on a dataset of quarterly earnings calls transcripts of companies in various sectors of the economy, as well as on a corpus of more recent earnings calls of companies in the S&P 1500 Composite Index. We find that our AMR-based approach outperforms fine-tuning LLMs directly on text in predicting stock price movement trends at different time horizons in both datasets. Our work also outperforms previous work utilizing document graphs and GNNs for text classification.
Authors: Mehdi Ali, Michael Fromm, Klaudia Thellmann, Jan Ebert, Alexander Arno Weber, Richard Rutmann, Charvi Jain, Max L\"ubbering, Daniel Steinigen, Johannes Leveling, Katrin Klug, Jasper Schulze Buschhoff, Lena Jurkschat, Hammam Abdelwahab, Benny J\"org Stein, Karl-Heinz Sylla, Pavel Denisov, Nicolo' Brandizzi, Qasid Saleem, Anirban Bhowmick, Lennard Helmer, Chelsea John, Pedro Ortiz Suarez, Malte Ostendorff, Alex Jude, Lalith Manjunath, Samuel Weinbach, Carolin Penke, Oleg Filatov, Shima Asaadi, Fabio Barth, Rafet Sifa, Fabian K\"uch, Andreas Herten, Ren\'e J\"akel, Georg Rehm, Stefan Kesselheim, Joachim K\"ohler, Nicolas Flores-Herr
Abstract: We present two multilingual LLMs designed to embrace Europe's linguistic diversity by supporting all 24 official languages of the European Union. Trained on a dataset comprising around 60% non-English data and utilizing a custom multilingual tokenizer, our models address the limitations of existing LLMs that predominantly focus on English or a few high-resource languages. We detail the models' development principles, i.e., data composition, tokenizer optimization, and training methodologies. The models demonstrate competitive performance across multilingual benchmarks, as evidenced by their performance on European versions of ARC, HellaSwag, MMLU, and TruthfulQA.
Authors: Amirhosein Ghasemabadi, Muhammad Kamran Janjua, Mohammad Salameh, Di Niu
Abstract: One key challenge to video restoration is to model the transition dynamics of video frames governed by motion. In this work, we propose TURTLE to learn the truncated causal history model for efficient and high-performing video restoration. Unlike traditional methods that process a range of contextual frames in parallel, TURTLE enhances efficiency by storing and summarizing a truncated history of the input frame latent representation into an evolving historical state. This is achieved through a sophisticated similarity-based retrieval mechanism that implicitly accounts for inter-frame motion and alignment. The causal design in TURTLE enables recurrence in inference through state-memorized historical features while allowing parallel training by sampling truncated video clips. We report new state-of-the-art results on a multitude of video restoration benchmark tasks, including video desnowing, nighttime video deraining, video raindrops and rain streak removal, video super-resolution, real-world and synthetic video deblurring, and blind video denoising while reducing the computational cost compared to existing best contextual methods on all these tasks.
Authors: Hadi Hosseini, Sanjukta Roy, Duohan Zhang
Abstract: Two-sided matching markets describe a large class of problems wherein participants from one side of the market must be matched to those from the other side according to their preferences. In many real-world applications (e.g. content matching or online labor markets), the knowledge about preferences may not be readily available and must be learned, i.e., one side of the market (aka agents) may not know their preferences over the other side (aka arms). Recent research on online settings has focused primarily on welfare optimization aspects (i.e. minimizing the overall regret) while paying little attention to the game-theoretic properties such as the stability of the final matching. In this paper, we exploit the structure of stable solutions to devise algorithms that improve the likelihood of finding stable solutions. We initiate the study of the sample complexity of finding a stable matching, and provide theoretical bounds on the number of samples needed to reach a stable matching with high probability. Finally, our empirical results demonstrate intriguing tradeoffs between stability and optimality of the proposed algorithms, further complementing our theoretical findings.
Authors: Yihuai Hong, Yuelin Zou, Lijie Hu, Ziqian Zeng, Di Wang, Haiqin Yang
Abstract: Fine-tuning-based unlearning methods prevail for preventing targeted harmful, sensitive, or copyrighted information within large language models while preserving overall capabilities. However, the true effectiveness of these methods is unclear. In this work, we delve into the limitations of fine-tuning-based unlearning through activation patching and parameter restoration experiments. Our findings reveal that these methods alter the model's knowledge retrieval process, providing further evidence that they do not genuinely erase the problematic knowledge embedded in the model parameters. Instead, the coefficients generated by the MLP components in the model's final layer are the primary contributors to these seemingly positive unlearning effects, playing a crucial role in controlling the model's behaviors. Furthermore, behavioral tests demonstrate that this unlearning mechanism inevitably impacts the global behavior of the models, affecting unrelated knowledge or capabilities. The code is released at https://github.com/yihuaihong/Dissecting-FT-Unlearning.
URLs: https://github.com/yihuaihong/Dissecting-FT-Unlearning.
Authors: Zeyu Zhang, Sixu Yan, Muzhi Han, Zaijin Wang, Xinggang Wang, Song-Chun Zhu, Hangxin Liu
Abstract: We propose M^3Bench, a new benchmark of whole-body motion generation for mobile manipulation tasks. Given a 3D scene context, M^3Bench requires an embodied agent to understand its configuration, environmental constraints and task objectives, then generate coordinated whole-body motion trajectories for object rearrangement tasks. M^3Bench features 30k object rearrangement tasks across 119 diverse scenes, providing expert demonstrations generated by our newly developed M^3BenchMaker. This automatic data generation tool produces coordinated whole-body motion trajectories from high-level task instructions, requiring only basic scene and robot information. Our benchmark incorporates various task splits to assess generalization across different dimensions and leverages realistic physics simulation for trajectory evaluation. Through extensive experimental analyses, we reveal that state-of-the-art models still struggle with coordinated base-arm motion while adhering to environment-context and task-specific constraints, highlighting the need to develop new models that address this gap. Through M^3Bench, we aim to facilitate future robotics research towards more adaptive and capable mobile manipulation in diverse, real-world environments.
Authors: Jake Callahan, Kevin Monogue, Ruben Villarreal, Tommie Catanach
Abstract: Monitoring networks increasingly aim to assimilate data from a large number of diverse sensors covering many sensing modalities. Bayesian optimal experimental design (OED) seeks to identify data, sensor configurations, or experiments which can optimally reduce uncertainty and hence increase the performance of a monitoring network. Information theory guides OED by formulating the choice of experiment or sensor placement as an optimization problem that maximizes the expected information gain (EIG) about quantities of interest given prior knowledge and models of expected observation data. Therefore, within the context of seismo-acoustic monitoring, we can use Bayesian OED to configure sensor networks by choosing sensor locations, types, and fidelity in order to improve our ability to identify and locate seismic sources. In this work, we develop the framework necessary to use Bayesian OED to optimize a sensor network's ability to locate seismic events from arrival time data of detected seismic phases at the regional-scale. Bayesian OED requires four elements: 1) A likelihood function that describes the distribution of detection and travel time data from the sensor network, 2) A Bayesian solver that uses a prior and likelihood to identify the posterior distribution of seismic events given the data, 3) An algorithm to compute EIG about seismic events over a dataset of hypothetical prior events, 4) An optimizer that finds a sensor network which maximizes EIG. Once we have developed this framework, we explore many relevant questions to monitoring such as: how to trade off sensor fidelity and earth model uncertainty; how sensor types, number, and locations influence uncertainty; and how prior models and constraints influence sensor placement.
Authors: Yu Fei, Yasaman Razeghi, Sameer Singh
Abstract: Large language models (LLMs) require alignment, such as instruction-tuning or reinforcement learning from human feedback, to effectively and safely follow user instructions. This process necessitates training aligned versions for every model size in each model family, resulting in significant computational overhead. In this work, we propose nudging, a simple, plug-and-play, and training-free algorithm that aligns any base model at inference time using a small aligned model. Nudging is motivated by recent findings that alignment primarily alters the model's behavior on a small subset of stylistic tokens, such as "Sure" or "Thank". We find that base models are significantly more uncertain when generating these tokens. Leveraging this observation, nudging employs a small aligned model to generate nudging tokens to steer the large base model's output toward desired directions when the base model's uncertainty is high. We evaluate the effectiveness of nudging across 3 model families and 13 tasks, covering reasoning, general knowledge, instruction following, and safety benchmarks. Without any additional training, nudging a large base model with a 7x - 14x smaller aligned model achieves zero-shot performance comparable to, and sometimes surpassing, that of large aligned models. For example, nudging OLMo-7b with OLMo-1b-instruct, affecting less than 9% of tokens, achieves a 10% absolute improvement on GSM8K over OLMo-7b-instruct. Unlike prior inference-time tuning methods, nudging enables off-the-shelf collaboration between model families. For instance, nudging Gemma-2-27b with Llama-2-7b-chat outperforms Llama-2-70b-chat on various tasks. Overall, this work introduces a simple yet powerful approach to token-level model collaboration, offering a modular solution to LLM alignment. Our project website: https://fywalter.github.io/nudging/ .
Authors: Kexin Li, Luwei Bai, Xiao Wang, Hao Wang
Abstract: Anderson acceleration is an effective technique for enhancing the efficiency of fixed-point iterations; however, analyzing its convergence in nonsmooth settings presents significant challenges. In this paper, we investigate a class of nonsmooth optimization algorithms characterized by the active manifold identification property. This class includes a diverse array of methods such as the proximal point method, proximal gradient method, proximal linear method, proximal coordinate descent method, Douglas-Rachford splitting (or the alternating direction method of multipliers), and the iteratively reweighted $\ell_1$ method, among others. Under the assumption that the optimization problem possesses an active manifold at a stationary point, we establish a local R-linear convergence rate for the Anderson-accelerated algorithm. Our extensive numerical experiments further highlight the robust performance of the proposed Anderson-accelerated methods.
Authors: Christopher Diehl, Peter Karkus, Sushant Veer, Marco Pavone, Torsten Bertram
Abstract: Distribution shifts between operational domains can severely affect the performance of learned models in self-driving vehicles (SDVs). While this is a well-established problem, prior work has mostly explored naive solutions such as fine-tuning, focusing on the motion prediction task. In this work, we explore novel adaptation strategies for differentiable autonomy stacks consisting of prediction, planning, and control, perform evaluation in closed-loop, and investigate the often-overlooked issue of catastrophic forgetting. Specifically, we introduce two simple yet effective techniques: a low-rank residual decoder (LoRD) and multi-task fine-tuning. Through experiments across three models conducted on two real-world autonomous driving datasets (nuPlan, exiD), we demonstrate the effectiveness of our methods and highlight a significant performance gap between open-loop and closed-loop evaluation in prior approaches. Our approach improves forgetting by up to 23.33% and the closed-loop OOD driving score by 8.83% in comparison to standard fine-tuning.
Authors: Achint Soni, Sreyas Venkataraman, Abhranil Chandra, Sebastian Fischmeister, Percy Liang, Bo Dai, Sherry Yang
Abstract: Video generation has been used to generate visual plans for controlling robotic systems. Given an image observation and a language instruction, previous work has generated video plans which are then converted to robot controls to be executed. However, a major bottleneck in leveraging video generation for control lies in the quality of the generated videos, which often suffer from hallucinatory content and unrealistic physics, resulting in low task success when control actions are extracted from the generated videos. While scaling up dataset and model size provides a partial solution, integrating external feedback is both natural and essential for grounding video generation in the real world. With this observation, we propose VideoAgent for self-improving generated video plans based on external feedback. Instead of directly executing the generated video plan, VideoAgent first refines the generated video plans using a novel procedure which we call self-conditioning consistency, utilizing feedback from a pretrained vision-language model (VLM). As the refined video plan is being executed, VideoAgent collects additional data from the environment to further improve video plan generation. Experiments in simulated robotic manipulation from MetaWorld and iTHOR show that VideoAgent drastically reduces hallucination, thereby boosting success rate of downstream manipulation tasks. We further illustrate that VideoAgent can effectively refine real-robot videos, providing an early indicator that robotics can be an effective tool in grounding video generation in the physical world.
Authors: Mu Cai, Reuben Tan, Jianrui Zhang, Bocheng Zou, Kai Zhang, Feng Yao, Fangrui Zhu, Jing Gu, Yiwu Zhong, Yuzhang Shang, Yao Dou, Jaden Park, Jianfeng Gao, Yong Jae Lee, Jianwei Yang
Abstract: Understanding fine-grained temporal dynamics is crucial for multimodal video comprehension and generation. Due to the lack of fine-grained temporal annotations, existing video benchmarks mostly resemble static image benchmarks and are incompetent at evaluating models for temporal understanding. In this paper, we introduce TemporalBench, a new benchmark dedicated to evaluating fine-grained temporal understanding in videos. TemporalBench consists of ~10K video question-answer pairs, derived from ~2K high-quality human annotations detailing the temporal dynamics in video clips. As a result, our benchmark provides a unique testbed for evaluating various temporal understanding and reasoning abilities such as action frequency, motion magnitude, event order, etc. Moreover, it enables evaluations on various tasks like both video question answering and captioning, both short and long video understanding, as well as different models such as multimodal video embedding models and text generation models. Results show that state-of-the-art models like GPT-4o achieve only 38.5% question answering accuracy on TemporalBench, demonstrating a significant gap (~30%) between humans and AI in temporal understanding. Furthermore, we notice a critical pitfall for multi-choice QA where LLMs can detect the subtle changes in negative captions and find a centralized description as a cue for its prediction, where we propose Multiple Binary Accuracy (MBA) to correct such bias. We hope that TemporalBench can foster research on improving models' temporal reasoning capabilities. Both dataset and evaluation code will be made available.