Authors: Haixiang Wu
Abstract: Current Transformer methods for Multivariate Time-Series Forecasting (MTSF) are all based on the conventional attention mechanism. They involve sequence embedding and performing a linear projection of Q, K, and V, and then computing attention within this latent space. We have never delved into the attention mechanism to explore whether such a mapping space is optimal for MTSF. To investigate this issue, this study first proposes Frequency Spectrum attention (FSatten), a novel attention mechanism based on the frequency domain space. It employs the Fourier transform for embedding and introduces Multi-head Spectrum Scaling (MSS) to replace the conventional linear mapping of Q and K. FSatten can accurately capture the periodic dependencies between sequences and outperform the conventional attention without changing mainstream architectures. We further design a more general method dubbed Scaled Orthogonal attention (SOatten). We propose an orthogonal embedding and a Head-Coupling Convolution (HCC) based on the neighboring similarity bias to guide the model in learning comprehensive dependency patterns. Experiments show that FSatten and SOatten surpass the SOTA which uses conventional attention, making it a good alternative as a basic attention mechanism for MTSF. The codes and log files will be released at: https://github.com/Joeland4/FSatten-SOatten.
Authors: Riddhiman Raut, Amit Kumar Ball, Amrita Basak
Abstract: This study presents novel predictive models using Graph Neural Networks (GNNs) for simulating thermal dynamics in Laser Powder Bed Fusion (L-PBF) processes. By developing and validating Single-Laser GNN (SL-GNN) and Multi-Laser GNN (ML-GNN) surrogates, the research introduces a scalable data-driven approach that learns fundamental physics from small-scale Finite Element Analysis (FEA) simulations and applies them to larger domains. Achieving a Mean Absolute Percentage Error (MAPE) of 3.77% with the baseline SL-GNN model, GNNs effectively learn from high-resolution simulations and generalize well across larger geometries. The proposed models capture the complexity of the heat transfer process in L-PBF while significantly reducing computational costs. For example, a thermomechanical simulation for a 2 mm x 2 mm domain typically requires about 4 hours, whereas the SL-GNN model can predict thermal distributions almost instantly. Calibrating models to larger domains enhances predictive performance, with significant drops in MAPE for 3 mm x 3 mm and 4 mm x 4 mm domains, highlighting the scalability and efficiency of this approach. Additionally, models show a decreasing trend in Root Mean Square Error (RMSE) when tuned to larger domains, suggesting potential for becoming geometry-agnostic. The interaction of multiple lasers complicates heat transfer, necessitating larger model architectures and advanced feature engineering. Using hyperparameters from Gaussian process-based Bayesian optimization, the best ML-GNN model demonstrates a 46.4% improvement in MAPE over the baseline ML-GNN model. In summary, this approach enables more efficient and flexible predictive modeling in L-PBF additive manufacturing.
Authors: Abdallah Alabdallah, Omar Hamed, Mattias Ohlsson, Thorsteinn R\"ognvaldsson, Sepideh Pashami
Abstract: The Cox Proportional Hazards (CPH) model has long been the preferred survival model for its explainability. However, to increase its predictive power beyond its linear log-risk, it was extended to utilize deep neural networks sacrificing its explainability. In this work, we explore the potential of self-explaining neural networks (SENN) for survival analysis. we propose a new locally explainable Cox proportional hazards model, named CoxSE, by estimating a locally-linear log-hazard function using the SENN. We also propose a modification to the Neural additive (NAM) models hybrid with SENN, named CoxSENAM, which enables the control of the stability and consistency of the generated explanations. Several experiments using synthetic and real datasets have been performed comparing with a NAM-based model, DeepSurv model explained with SHAP, and a linear CPH model. The results show that, unlike the NAM-based model, the SENN-based model can provide more stable and consistent explanations while maintaining the same expressiveness power of the black-box model. The results also show that, due to their structural design, NAM-based models demonstrated better robustness to non-informative features. Among these models, the hybrid model exhibited the best robustness.
Authors: Seonho Lee, Amar Phanishayee, Divya Mahajan
Abstract: Deep learning kernels exhibit predictable memory accesses and compute patterns, making GPUs' parallel architecture well-suited for their execution. Software and runtime systems for GPUs are optimized to better utilize the stream multiprocessors, on-chip cache, and off-chip high-bandwidth memory. As deep learning models and GPUs evolve, access to newer GPUs is often limited, raising questions about the performance of new model architectures on existing GPUs, existing models on new GPUs, and new model architectures on new GPUs. To address these questions, we introduce NeuSight, a framework to predict the performance of various deep learning models, for both training and inference, on unseen GPUs without requiring actual execution. The framework leverages both GPU hardware behavior and software library optimizations to estimate end-to-end performance. Previous work uses regression models that capture linear trends or multilayer perceptrons to predict the overall latency of deep learning kernels on GPUs. These approaches suffer from higher error percentages when forecasting performance on unseen models and new GPUs. Instead, NeuSight decomposes the prediction problem into smaller problems, bounding the prediction through fundamental performance laws. NeuSight decomposes a single deep learning kernel prediction into smaller working sets called tiles, which are executed independently on the GPU. Tile-granularity predictions are determined using a machine learning approach and aggregated to estimate end-to-end latency. NeuSight outperforms prior work across various deep learning workloads and the latest GPUs. It reduces the percentage error from 198% and 19.7% to 3.8% in predicting the latency of GPT3 model for training and inference on H100, compared to state-of-the-art prior works, where both GPT3 and H100 were not used to train the framework.
Authors: Janis Adamek, Moritz Schulze Darup
Abstract: Federated learning (FL) schemes allow multiple participants to collaboratively train neural networks without the need to directly share the underlying data.However, in early schemes, all participants eventually obtain the same model. Moreover, the aggregation is typically carried out by a third party, who obtains combined gradients or weights, which may reveal the model. These downsides underscore the demand for fair and privacy-preserving FL schemes. Here, collaborative fairness asks for individual model quality depending on the individual data contribution. Privacy is demanded with respect to any kind of data outsourced to the third party. Now, there already exist some approaches aiming for either fair or privacy-preserving FL and a few works even address both features. In our paper, we build upon these seminal works and present a novel, fair and privacy-preserving FL scheme. Our approach, which mainly relies on homomorphic encryption, stands out for exclusively using local gradients. This increases the usability in comparison to state-of-the-art approaches and thereby opens the door to applications in control.
Authors: Moritz Th\"uning
Abstract: When implementations of the Transformer's self-attention layer utilize SRAM instead of DRAM, they can achieve significant speedups. The Tenstorrent Grayskull architecture provides a large SRAM, distributed across a grid of cores. This work presents a fused kernel for Grayskull, that exclusively utilizes its large SRAM by combining matrix multiplication, attention score scaling and Softmax operations. Additionally, a dedicated Softmax kernel utilizing the SRAM and a CPU implementation serving as a baseline are presented. The Softmax operation consumes most of the runtime in the computation of attention weights from queries and keys on Grayskull. The speedup of the dedicated Softmax kernel compared to the CPU implementation is up to $10 \times$, and the Softmax implementation inside the fused kernel is approximately $1.8 \times$ faster than the dedicated Softmax kernel. The time and memory complexity of all implementations is quadratic in sequence length. Currently, the Grayskull e150 is approximately $30 \times$ cheaper for the general public than an Nvidia H100 PCIe (a state-of-the-art GPU) and offers approximately $1.5 \times$ more SRAM.
Authors: Yi Sheng, Junhuan Yang, Jinyang Li, James Alaina, Xiaowei Xu, Yiyu Shi, Jingtong Hu, Weiwen Jiang, Lei Yang
Abstract: As Artificial Intelligence (AI) increasingly integrates into our daily lives, fairness has emerged as a critical concern, particularly in medical AI, where datasets often reflect inherent biases due to social factors like the underrepresentation of marginalized communities and socioeconomic barriers to data collection. Traditional approaches to mitigating these biases have focused on data augmentation and the development of fairness-aware training algorithms. However, this paper argues that the architecture of neural networks, a core component of Machine Learning (ML), plays a crucial role in ensuring fairness. We demonstrate that addressing fairness effectively requires a holistic approach that simultaneously considers data, algorithms, and architecture. Utilizing Automated ML (AutoML) technology, specifically Neural Architecture Search (NAS), we introduce a novel framework, BiaslessNAS, designed to achieve fair outcomes in analyzing skin lesion datasets. BiaslessNAS incorporates fairness considerations at every stage of the NAS process, leading to the identification of neural networks that are not only more accurate but also significantly fairer. Our experiments show that BiaslessNAS achieves a 2.55% increase in accuracy and a 65.50% improvement in fairness compared to traditional NAS methods, underscoring the importance of integrating fairness into neural network architecture for better outcomes in medical AI applications.
Authors: Nasir Khan, Sinem Coleri
Abstract: Future 6G-enabled vehicular networks face the challenge of ensuring ultra-reliable low-latency communication (URLLC) for delivering safety-critical information in a timely manner. Existing resource allocation schemes for vehicle-to-everything (V2X) communication systems primarily rely on traditional optimization-based algorithms. However, these methods often fail to guarantee the strict reliability and latency requirements of URLLC applications in dynamic vehicular environments due to the high complexity and communication overhead of the solution methodologies. This paper proposes a novel deep reinforcement learning (DRL) based framework for the joint power and block length allocation to minimize the worst-case decoding-error probability in the finite block length (FBL) regime for a URLLC-based downlink V2X communication system. The problem is formulated as a non-convex mixed-integer nonlinear programming problem (MINLP). Initially, an algorithm grounded in optimization theory is developed based on deriving the joint convexity of the decoding error probability in the block length and transmit power variables within the region of interest. Subsequently, an efficient event-triggered DRL-based algorithm is proposed to solve the joint optimization problem. Incorporating event-triggered learning into the DRL framework enables assessing whether to initiate the DRL process, thereby reducing the number of DRL process executions while maintaining reasonable reliability performance. Simulation results demonstrate that the proposed event-triggered DRL scheme can achieve 95% of the performance of the joint optimization scheme while reducing the DRL executions by up to 24% for different network settings.
Authors: Muhammad Uzair Zahid, Serkan Kiranyaz, Alper Yildirim, Moncef Gabbouj
Abstract: Objective: Many studies on radar signal restoration in the literature focus on isolated restoration problems, such as denoising over a certain type of noise, while ignoring other types of artifacts. Additionally, these approaches usually assume a noisy environment with a limited set of fixed signal-to-noise ratio (SNR) levels. However, real-world radar signals are often corrupted by a blend of artifacts, including but not limited to unwanted echo, sensor noise, intentional jamming, and interference, each of which can vary in type, severity, and duration. This study introduces Blind Radar Signal Restoration using an Operational Generative Adversarial Network (BRSR-OpGAN), which uses a dual domain loss in the temporal and spectral domains. This approach is designed to improve the quality of radar signals, regardless of the diversity and intensity of the corruption. Methods: The BRSR-OpGAN utilizes 1D Operational GANs, which use a generative neuron model specifically optimized for blind restoration of corrupted radar signals. This approach leverages GANs' flexibility to adapt dynamically to a wide range of artifact characteristics. Results: The proposed approach has been extensively evaluated using a well-established baseline and a newly curated extended dataset called the Blind Radar Signal Restoration (BRSR) dataset. This dataset was designed to simulate real-world conditions and includes a variety of artifacts, each varying in severity. The evaluation shows an average SNR improvement over 15.1 dB and 14.3 dB for the baseline and BRSR datasets, respectively. Finally, even on resource-constrained platforms, the proposed approach can be applied in real-time.
Authors: Suryanarayanan Manoj Sanu, Alejandro M. Aragon, Miguel A. Bessa
Abstract: Neural networks (NNs) hold great promise for advancing inverse design via topology optimization (TO), yet misconceptions about their application persist. This article focuses on neural topology optimization (neural TO), which leverages NNs to reparameterize the decision space and reshape the optimization landscape. While the method is still in its infancy, our analysis tools reveal critical insights into the NNs' impact on the optimization process. We demonstrate that the choice of NN architecture significantly influences the objective landscape and the optimizer's path to an optimum. Notably, NNs introduce non-convexities even in otherwise convex landscapes, potentially delaying convergence in convex problems but enhancing exploration for non-convex problems. This analysis lays the groundwork for future advancements by highlighting: 1) the potential of neural TO for non-convex problems and dedicated GPU hardware (the "good"), 2) the limitations in smooth landscapes (the "bad"), and 3) the complex challenge of selecting optimal NN architectures and hyperparameters for superior performance (the "ugly").
Authors: Tyler LaBonte, John C. Hill, Xinchen Zhang, Vidya Muthukumar, Abhishek Kumar
Abstract: Modern machine learning models are prone to over-reliance on spurious correlations, which can often lead to poor performance on minority groups. In this paper, we identify surprising and nuanced behavior of finetuned models on worst-group accuracy via comprehensive experiments on four well-established benchmarks across vision and language tasks. We first show that the commonly used class-balancing techniques of mini-batch upsampling and loss upweighting can induce a decrease in worst-group accuracy (WGA) with training epochs, leading to performance no better than without class-balancing. While in some scenarios, removing data to create a class-balanced subset is more effective, we show this depends on group structure and propose a mixture method which can outperform both techniques. Next, we show that scaling pretrained models is generally beneficial for worst-group accuracy, but only in conjuction with appropriate class-balancing. Finally, we identify spectral imbalance in finetuning features as a potential source of group disparities -- minority group covariance matrices incur a larger spectral norm than majority groups once conditioned on the classes. Our results show more nuanced interactions of modern finetuned models with group robustness than was previously known. Our code is available at https://github.com/tmlabonte/revisiting-finetuning.
Authors: Guangqiang Li, M. Amine Atoui, Xiangshun Li
Abstract: Domain generalization achieves fault diagnosis on unseen modes. In process industrial systems, fault samples are limited, and only single-mode fault data can be obtained. Extracting domain-invariant fault features from single-mode data for unseen mode fault diagnosis poses challenges. Existing methods utilize a generator module to simulate samples of unseen modes. However, multi-mode samples contain complex spatiotemporal information, which brings significant difficulties to accurate sample generation. Therefore, double gradient reversal network (DGRN) is proposed. First, the model is pre-trained to acquire fault knowledge from the single seen mode. Then, pseudo-fault feature generation strategy is designed by Adaptive instance normalization, to simulate fault features of unseen mode. The dual adversarial training strategy is created to enhance the diversity of pseudo-fault features, which models unseen modes with significant distribution differences. Subsequently, domain-invariant feature extraction strategy is constructed by contrastive learning and adversarial learning. This strategy extracts common features of faults and helps multi-mode fault diagnosis. Finally, the experiments were conducted on Tennessee Eastman process and continuous stirred-tank reactor. The experiments demonstrate that DGRN achieves high classification accuracy on unseen modes while maintaining a small model size.
Authors: Nika Haghtalab, Mingda Qiao, Kunhe Yang, Eric Zhao
Abstract: We initiate the study of the truthfulness of calibration measures in sequential prediction. A calibration measure is said to be truthful if the forecaster (approximately) minimizes the expected penalty by predicting the conditional expectation of the next outcome, given the prior distribution of outcomes. Truthfulness is an important property of calibration measures, ensuring that the forecaster is not incentivized to exploit the system with deliberate poor forecasts. This makes it an essential desideratum for calibration measures, alongside typical requirements, such as soundness and completeness. We conduct a taxonomy of existing calibration measures and their truthfulness. Perhaps surprisingly, we find that all of them are far from being truthful. That is, under existing calibration measures, there are simple distributions on which a polylogarithmic (or even zero) penalty is achievable, while truthful prediction leads to a polynomial penalty. Our main contribution is the introduction of a new calibration measure termed the Subsampled Smooth Calibration Error (SSCE) under which truthful prediction is optimal up to a constant multiplicative factor.
Authors: Quan Li, Tianxiang Zhao, Lingwei Chen, Junjie Xu, Suhang Wang
Abstract: Graphs have emerged as critical data structures for content analysis in various domains, such as social network analysis, bioinformatics, and recommendation systems. Node classification, a fundamental task in this context, is typically tackled using graph neural networks (GNNs). Unfortunately, conventional GNNs still face challenges in scenarios with few labeled nodes, despite the prevalence of few-shot node classification tasks in real-world applications. To address this challenge, various approaches have been proposed, including graph meta-learning, transfer learning, and methods based on Large Language Models (LLMs). However, traditional meta-learning and transfer learning methods often require prior knowledge from base classes or fail to exploit the potential advantages of unlabeled nodes. Meanwhile, LLM-based methods may overlook the zero-shot capabilities of LLMs and rely heavily on the quality of generated contexts. In this paper, we propose a novel approach that integrates LLMs and GNNs, leveraging the zero-shot inference and reasoning capabilities of LLMs and employing a Graph-LLM-based active learning paradigm to enhance GNNs' performance. Extensive experiments demonstrate the effectiveness of our model in improving node classification accuracy with considerably limited labeled data, surpassing state-of-the-art baselines by significant margins.
Authors: Danielle Ensign, Adri\`a Garriga-Alonso
Abstract: How well will current interpretability techniques generalize to future models? A relevant case study is Mamba, a recent recurrent architecture with scaling comparable to Transformers. We adapt pre-Mamba techniques to Mamba and partially reverse-engineer the circuit responsible for the Indirect Object Identification (IOI) task. Our techniques provide evidence that 1) Layer 39 is a key bottleneck, 2) Convolutions in layer 39 shift names one position forward, and 3) The name entities are stored linearly in Layer 39's SSM. Finally, we adapt an automatic circuit discovery tool, positional Edge Attribution Patching, to identify a Mamba IOI circuit. Our contributions provide initial evidence that circuit-based mechanistic interpretability tools work well for the Mamba architecture.
Authors: Dong Hyeon Mok, Seoin Back
Abstract: Discovery of novel and promising materials is a critical challenge in the field of chemistry and material science, traditionally approached through methodologies ranging from trial-and-error to machine learning-driven inverse design. Recent studies suggest that transformer-based language models can be utilized as material generative models to expand chemical space and explore materials with desired properties. In this work, we introduce the Catalyst Generative Pretrained Transformer (CatGPT), trained to generate string representations of inorganic catalyst structures from a vast chemical space. CatGPT not only demonstrates high performance in generating valid and accurate catalyst structures but also serves as a foundation model for generating desired types of catalysts by fine-tuning with sparse and specified datasets. As an example, we fine-tuned the pretrained CatGPT using a binary alloy catalyst dataset designed for screening two-electron oxygen reduction reaction (2e-ORR) catalyst and generate catalyst structures specialized for 2e-ORR. Our work demonstrates the potential of language models as generative tools for catalyst discovery.
Authors: Jingyao Wang, Wenwen Qiang, Jiangmeng Li, Lingyu Si, Changwen Zheng, Bing Su
Abstract: An effective paradigm of multi-modal learning (MML) is to learn unified representations among modalities. From a causal perspective, constraining the consistency between different modalities can mine causal representations that convey primary events. However, such simple consistency may face the risk of learning insufficient or unnecessary information: a necessary but insufficient cause is invariant across modalities but may not have the required accuracy; a sufficient but unnecessary cause tends to adapt well to specific modalities but may be hard to adapt to new data. To address this issue, in this paper, we aim to learn representations that are both causal sufficient and necessary, i.e., Causal Complete Cause ($C^3$), for MML. Firstly, we define the concept of $C^3$ for MML, which reflects the probability of being causal sufficiency and necessity. We also propose the identifiability and measurement of $C^3$, i.e., $C^3$ risk, to ensure calculating the learned representations' $C^3$ scores in practice. Then, we theoretically prove the effectiveness of $C^3$ risk by establishing the performance guarantee of MML with a tight generalization bound. Based on these theoretical results, we propose a plug-and-play method, namely Causal Complete Cause Regularization ($C^3$R), to learn causal complete representations by constraining the $C^3$ risk bound. Extensive experiments conducted on various benchmark datasets empirically demonstrate the effectiveness of $C^3$R.
Authors: Shuang Li, Ziyuan Pu, Nan Zhang, Duxin Chen, Lu Dong, Daniel J. Graham, Yinhai Wang
Abstract: Traffic crashes profoundly impede traffic efficiency and pose economic challenges. Accurate prediction of post-crash traffic status provides essential information for evaluating traffic perturbations and developing effective solutions. Previous studies have established a series of deep learning models to predict post-crash traffic conditions, however, these correlation-based methods cannot accommodate the biases caused by time-varying confounders and the heterogeneous effects of crashes. The post-crash traffic prediction model needs to estimate the counterfactual traffic speed response to hypothetical crashes under various conditions, which demonstrates the necessity of understanding the causal relationship between traffic factors. Therefore, this paper presents the Marginal Structural Causal Transformer (MSCT), a novel deep learning model designed for counterfactual post-crash traffic prediction. To address the issue of time-varying confounding bias, MSCT incorporates a structure inspired by Marginal Structural Models and introduces a balanced loss function to facilitate learning of invariant causal features. The proposed model is treatment-aware, with a specific focus on comprehending and predicting traffic speed under hypothetical crash intervention strategies. In the absence of ground-truth data, a synthetic data generation procedure is proposed to emulate the causal mechanism between traffic speed, crashes, and covariates. The model is validated using both synthetic and real-world data, demonstrating that MSCT outperforms state-of-the-art models in multi-step-ahead prediction performance. This study also systematically analyzes the impact of time-varying confounding bias and dataset distribution on model performance, contributing valuable insights into counterfactual prediction for intelligent transportation systems.
Authors: Tobias Kerner
Abstract: There are many cases where LLMs are used for specific tasks in a single domain. These usually require less general, but more domain-specific knowledge. Highly capable, general-purpose state-of-the-art language models like GPT-4 or Claude-3-opus can often be used for such tasks, but they are very large and cannot be run locally, even if they were not proprietary. This can be a problem when working with sensitive data. This paper focuses on domain-specific and mixed-domain pretraining as potentially more efficient methods than general pretraining for specialized language models. We will take a look at work related to domain-specific pretraining, specifically in the medical area, and compare benchmark results of specialized language models to general-purpose language models.
Authors: Yifan Wang, Xiao Luo, Chong Chen, Xian-Sheng Hua, Ming Zhang, Wei Ju
Abstract: Graph classification is a critical task in numerous multimedia applications, where graphs are employed to represent diverse types of multimedia data, including images, videos, and social networks. Nevertheless, in real-world scenarios, labeled graph data can be limited or scarce. To address this issue, we focus on the problem of semi-supervised graph classification, which involves both supervised and unsupervised models learning from labeled and unlabeled data. In contrast to recent approaches that transfer the entire knowledge from the unsupervised model to the supervised one, we argue that an effective transfer should only retain the relevant semantics that align well with the supervised task. In this paper, we propose a novel framework named DisenSemi, which learns disentangled representation for semi-supervised graph classification. Specifically, a disentangled graph encoder is proposed to generate factor-wise graph representations for both supervised and unsupervised models. Then we train two models via supervised objective and mutual information (MI)-based constraints respectively. To ensure the meaningful transfer of knowledge from the unsupervised encoder to the supervised one, we further define an MI-based disentangled consistency regularization between two models and identify the corresponding rationale that aligns well with the current graph classification task. Experimental results on a range of publicly accessible datasets reveal the effectiveness of our DisenSemi.
Authors: Erik B. Terres-Escudero, Javier Del Ser, Aitor Mart\'inez-Seras, Pablo Garcia-Bringas
Abstract: In the last decade, Artificial Intelligence (AI) models have rapidly integrated into production pipelines propelled by their excellent modeling performance. However, the development of these models has not been matched by advancements in algorithms ensuring their safety, failing to guarantee robust behavior against Out-of-Distribution (OoD) inputs outside their learning domain. Furthermore, there is a growing concern with the sustainability of AI models and their required energy consumption in both training and inference phases. To mitigate these issues, this work explores the use of the Forward-Forward Algorithm (FFA), a biologically plausible alternative to Backpropagation, adapted to the spiking domain to enhance the overall energy efficiency of the model. By capitalizing on the highly expressive topology emerging from the latent space of models trained with FFA, we develop a novel FF-SCP algorithm for OoD Detection. Our approach measures the likelihood of a sample belonging to the in-distribution (ID) data by using the distance from the latent representation of samples to class-representative manifolds. Additionally, to provide deeper insights into our OoD pipeline, we propose a gradient-free attribution technique that highlights the features of a sample pushing it away from the distribution of any class. Multiple experiments using our spiking FFA adaptation demonstrate that the achieved accuracy levels are comparable to those seen in analog networks trained via back-propagation. Furthermore, OoD detection experiments on multiple datasets prove that FF-SCP outperforms avant-garde OoD detectors within the spiking domain in terms of several metrics used in this area. We also present a qualitative analysis of our explainability technique, exposing the precision by which the method detects OoD features, such as embedded artifacts or missing regions.
Authors: Matthias Karlbauer, Danielle C. Maddix, Abdul Fatir Ansari, Boran Han, Gaurav Gupta, Yuyang Wang, Andrew Stuart, Michael W. Mahoney
Abstract: Remarkable progress in the development of Deep Learning Weather Prediction (DLWP) models positions them to become competitive with traditional numerical weather prediction (NWP) models. Indeed, a wide number of DLWP architectures -- based on various backbones, including U-Net, Transformer, Graph Neural Network (GNN), and Fourier Neural Operator (FNO) -- have demonstrated their potential at forecasting atmospheric states. However, due to differences in training protocols, forecast horizons, and data choices, it remains unclear which (if any) of these methods and architectures are most suitable for weather forecasting and for future model development. Here, we step back and provide a detailed empirical analysis, under controlled conditions, comparing and contrasting the most prominent DLWP models, along with their backbones. We accomplish this by predicting synthetic two-dimensional incompressible Navier-Stokes and real-world global weather dynamics. In terms of accuracy, memory consumption, and runtime, our results illustrate various tradeoffs. For example, on synthetic data, we observe favorable performance of FNO; and on the real-world WeatherBench dataset, our results demonstrate the suitability of ConvLSTM and SwinTransformer for short-to-mid-ranged forecasts. For long-ranged weather rollouts of up to 365 days, we observe superior stability and physical soundness in architectures that formulate a spherical data representation, i.e., GraphCast and Spherical FNO. In addition, we observe that all of these model backbones ``saturate,'' i.e., none of them exhibit so-called neural scaling, which highlights an important direction for future work on these and related models.
Authors: Neil De La Fuente, Daniel A. Vidal Guerra
Abstract: This study conducts a comparative analysis of three advanced Deep Reinforcement Learning models: Deep Q-Networks (DQN), Proximal Policy Optimization (PPO), and Advantage Actor-Critic (A2C), within the BreakOut Atari game environment. Our research assesses the performance and effectiveness of these models in a controlled setting. Through rigorous experimentation, we examine each model's learning efficiency, strategy development, and adaptability under dynamic game conditions. The findings provide critical insights into the practical applications of these models in game-based learning environments and contribute to the broader understanding of their capabilities. The code is publicly available at github.com/Neilus03/DRL_comparative_study.
Authors: Janez Bo\v{z}i\v{c}, Am\^andio R. Faustino, Boris Radovi\v{c}, Marco Canini, Veljko Pejovi\'c
Abstract: Progressing beyond centralized AI is of paramount importance, yet, distributed AI solutions, in particular various federated learning (FL) algorithms, are often not comprehensively assessed, which prevents the research community from identifying the most promising approaches and practitioners from being convinced that a certain solution is deployment-ready. The largest hurdle towards FL algorithm evaluation is the difficulty of conducting real-world experiments over a variety of FL client devices and different platforms, with different datasets and data distribution, all while assessing various dimensions of algorithm performance, such as inference accuracy, energy consumption, and time to convergence, to name a few. In this paper, we present CoLExT, a real-world testbed for FL research. CoLExT is designed to streamline experimentation with custom FL algorithms in a rich testbed configuration space, with a large number of heterogeneous edge devices, ranging from single-board computers to smartphones, and provides real-time collection and visualization of a variety of metrics through automatic instrumentation. According to our evaluation, porting FL algorithms to CoLExT requires minimal involvement from the developer, and the instrumentation introduces minimal resource usage overhead. Furthermore, through an initial investigation involving popular FL algorithms running on CoLExT, we reveal previously unknown trade-offs, inefficiencies, and programming bugs.
Authors: Paul M Baggenstoss
Abstract: We revisit the classical problem of inverting dimension-reducing linear mappings using the maximum entropy (MaxEnt) criterion. In the literature, solutions are problem-dependent, inconsistent, and use different entropy measures. We propose a new unified approach that not only specializes to the existing approaches, but offers solutions to new cases, such as when data values are constrained to [0, 1], which has new applications in machine learning.
Authors: Hannah Rosa Friesacher, Ola Engkvist, Lewis Mervin, Yves Moreau, Adam Arany
Abstract: In the drug discovery process, where experiments can be costly and time-consuming, computational models that predict drug-target interactions are valuable tools to accelerate the development of new therapeutic agents. Estimating the uncertainty inherent in these neural network predictions provides valuable information that facilitates optimal decision-making when risk assessment is crucial. However, such models can be poorly calibrated, which results in unreliable uncertainty estimates that do not reflect the true predictive uncertainty. In this study, we compare different metrics, including accuracy and calibration scores, used for model hyperparameter tuning to investigate which model selection strategy achieves well-calibrated models. Furthermore, we propose to use a computationally efficient Bayesian uncertainty estimation method named Bayesian Linear Probing (BLP), which generates Hamiltonian Monte Carlo (HMC) trajectories to obtain samples for the parameters of a Bayesian Logistic Regression fitted to the hidden layer of the baseline neural network. We report that BLP improves model calibration and achieves the performance of common uncertainty quantification methods by combining the benefits of uncertainty estimation and probability calibration methods. Finally, we show that combining post hoc calibration method with well-performing uncertainty quantification approaches can boost model accuracy and calibration.
Authors: Hongyan Chang, Hamed Hassani, Reza Shokri
Abstract: Watermarking is a technique used to embed a hidden signal in the probability distribution of text generated by large language models (LLMs), enabling attribution of the text to the originating model. We introduce smoothing attacks and show that existing watermarking methods are not robust against minor modifications of text. An adversary can use weaker language models to smooth out the distribution perturbations caused by watermarks without significantly compromising the quality of the generated text. The modified text resulting from the smoothing attack remains close to the distribution of text that the original model (without watermark) would have produced. Our attack reveals a fundamental limitation of a wide range of watermarking techniques.
Authors: Bo Liu, Rui Wang, Lemeng Wu, Yihao Feng, Peter Stone, Qiang Liu
Abstract: The most fundamental capability of modern AI methods such as Large Language Models (LLMs) is the ability to predict the next token in a long sequence of tokens, known as ``sequence modeling." Although the Transformers model is the current dominant approach to sequence modeling, its quadratic computational cost with respect to sequence length is a significant drawback. State-space models (SSMs) offer a promising alternative due to their linear decoding efficiency and high parallelizability during training. However, existing SSMs often rely on seemingly ad hoc linear recurrence designs. In this work, we explore SSM design through the lens of online learning, conceptualizing SSMs as meta-modules for specific online learning problems. This approach links SSM design to formulating precise online learning objectives, with state transition rules derived from optimizing these objectives. Based on this insight, we introduce a novel deep SSM architecture based on the implicit update for optimizing an online regression objective. Our experimental results show that our models outperform state-of-the-art SSMs, including the Mamba model, on standard sequence modeling benchmarks and language modeling tasks.
Authors: Shiqi Liu, Yihua Tan
Abstract: With the advancement of computer vision and natural language processing, text-to-video generation, enabled by text-to-video diffusion models, has become more prevalent. These models are trained using a large amount of data from the internet. However, the training data often contain copyrighted content, including cartoon character icons and artist styles, private portraits, and unsafe videos. Since filtering the data and retraining the model is challenging, methods for unlearning specific concepts from text-to-video diffusion models have been investigated. However, due to the high computational complexity and relative large optimization scale, there is little work on unlearning methods for text-to-video diffusion models. We propose a novel concept-unlearning method by transferring the unlearning capability of the text encoder of text-to-image diffusion models to text-to-video diffusion models. Specifically, the method optimizes the text encoder using few-shot unlearning, where several generated images are used. We then use the optimized text encoder in text-to-video diffusion models to generate videos. Our method costs low computation resources and has small optimization scale. We discuss the generated videos after unlearning a concept. The experiments demonstrates that our method can unlearn copyrighted cartoon characters, artist styles, objects and people's facial characteristics. Our method can unlearn a concept within about 100 seconds on an RTX 3070. Since there was no concept unlearning method for text-to-video diffusion models before, we make concept unlearning feasible and more accessible in the text-to-video domain.
Authors: Jos\'e Daniel Pascual-Triana, Alberto Fern\'andez, Paulo Novais, Francisco Herrera
Abstract: Given the magnitude of data generation currently, both in quantity and speed, the use of machine learning is increasingly important. When data include protected features that might give rise to discrimination, special care must be taken. Data quality is critical in these cases, as biases in training data can be reflected in classification models. This has devastating consequences and fails to comply with current regulations. Data-Centric Artificial Intelligence proposes dataset modifications to improve its quality. Instance selection via undersampling can foster balanced learning of classes and protected feature values in the classifier. When such undersampling is done close to the decision boundary, the effect on the classifier would be bolstered. This work proposes Fair Overlap Number of Balls (Fair-ONB), an undersampling method that harnesses the data morphology of the different data groups (obtained from the combination of classes and protected feature values) to perform guided undersampling in the areas where they overlap. It employs attributes of the ball coverage of the groups, such as the radius, number of covered instances and density, to select the most suitable areas for undersampling and reduce bias. Results show that the Fair-ONB method reduces bias with low impact on the classifier's predictive performance.
Authors: Armin Abdollahi, Xinghong Ma, Jiahao Zhang, Daijia Wu, Tongshou Wu, Zizheng Ye, Maryam Pishgar
Abstract: Background: Stroke is second-leading cause of disability and death among adults. Approximately 17 million people suffer from a stroke annually, with about 85% being ischemic strokes. Predicting mortality of ischemic stroke patients in intensive care unit (ICU) is crucial for optimizing treatment strategies, allocating resources, and improving survival rates. Methods: We acquired data on ICU ischemic stroke patients from MIMIC-IV database, including diagnoses, vital signs, laboratory tests, medications, procedures, treatments, and clinical notes. Stroke patients were randomly divided into training (70%, n=2441), test (15%, n=523), and validation (15%, n=523) sets. To address data imbalances, we applied Synthetic Minority Over-sampling Technique (SMOTE). We selected 30 features for model development, significantly reducing feature number from 1095 used in the best study. We developed a deep learning model to assess mortality risk and implemented several baseline machine learning models for comparison. Results: XGB-DL model, combining XGBoost for feature selection and deep learning, effectively minimized false positives. Model AUROC improved from 0.865 (95% CI: 0.821 - 0.905) on first day to 0.903 (95% CI: 0.868 - 0.936) by fourth day using data from 3,646 ICU mortality patients in the MIMIC-IV database with 0.945 AUROC (95% CI: 0.944 - 0.947) during training. Although other ML models also performed well in terms of AUROC, we chose Deep Learning for its higher specificity. Conclusions: Through enhanced feature selection and data cleaning, proposed model demonstrates a 13% AUROC improvement compared to existing models while reducing feature number from 1095 in previous studies to 30.
Authors: Chao Min, Guoquan Wen, Jiangru Yuan, Jun Yi, Xing Guo
Abstract: Industrial time-series, as a structural data responds to production process information, can be utilized to perform data-driven decision-making for effective monitoring of industrial production process. However, there are some challenges for time-series forecasting in industry, e.g., predicting few-shot caused by data shortage, and decision-confusing caused by unknown treatment policy. To cope with the problems, we propose a novel causal domain adaptation framework, Causal Domain Adaptation (CDA) forecaster to improve the performance on the interested domain with limited data (target). Firstly, we analyze the causality existing along with treatments, and thus ensure the shared causality over time. Subsequently, we propose an answer-based attention mechanism to achieve domain-invariant representation by the shared causality in both domains. Then, a novel domain-adaptation is built to model treatments and outcomes jointly training on source and target domain. The main insights are that our designed answer-based attention mechanism allows the target domain to leverage the existed causality in source time-series even with different treatments, and our forecaster can predict the counterfactual outcome of industrial time-series, meaning a guidance in production process. Compared with commonly baselines, our method on real-world and synthetic oilfield datasets demonstrates the effectiveness in across-domain prediction and the practicality in guiding production process
Authors: Sebastian Cygert, Damian S\'ojka, Tomasz Trzci\'nski, Bart{\l}omiej Twardowski
Abstract: Test-Time Adaptation (TTA) has recently emerged as a promising strategy for tackling the problem of machine learning model robustness under distribution shifts by adapting the model during inference without access to any labels. Because of task difficulty, hyperparameters strongly influence the effectiveness of adaptation. However, the literature has provided little exploration into optimal hyperparameter selection. In this work, we tackle this problem by evaluating existing TTA methods using surrogate-based hp-selection strategies (which do not assume access to the test labels) to obtain a more realistic evaluation of their performance. We show that some of the recent state-of-the-art methods exhibit inferior performance compared to the previous algorithms when using our more realistic evaluation setup. Further, we show that forgetting is still a problem in TTA as the only method that is robust to hp-selection resets the model to the initial state at every step. We analyze different types of unsupervised selection strategies, and while they work reasonably well in most scenarios, the only strategies that work consistently well use some kind of supervision (either by a limited number of annotated test samples or by using pretraining data). Our findings underscore the need for further research with more rigorous benchmarking by explicitly stating model selection strategies, to facilitate which we open-source our code.
Authors: Xianda Chen, PakHin Tiu, Xu Han, Junjie Chen, Yuanfei Wu, Xinhu Zheng, Meixin Zhu
Abstract: The continual evolution of autonomous driving technology requires car-following models that can adapt to diverse and dynamic traffic environments. Traditional learning-based models often suffer from performance degradation when encountering unseen traffic patterns due to a lack of continual learning capabilities. This paper proposes a novel car-following model based on continual learning that addresses this limitation. Our framework incorporates Elastic Weight Consolidation (EWC) and Memory Aware Synapses (MAS) techniques to mitigate catastrophic forgetting and enable the model to learn incrementally from new traffic data streams. We evaluate the performance of the proposed model on the Waymo and Lyft datasets which encompass various traffic scenarios. The results demonstrate that the continual learning techniques significantly outperform the baseline model, achieving 0\% collision rates across all traffic conditions. This research contributes to the advancement of autonomous driving technology by fostering the development of more robust and adaptable car-following models.
Authors: Sourasekhar Banerjee, Ali Dadras, Alp Yurtsever, Monowar Bhuyan
Abstract: The key challenge of personalized federated learning (PerFL) is to capture the statistical heterogeneity properties of data with inexpensive communications and gain customized performance for participating devices. To address these, we introduced personalized federated learning in multi-tier architecture (PerMFL) to obtain optimized and personalized local models when there are known team structures across devices. We provide theoretical guarantees of PerMFL, which offers linear convergence rates for smooth strongly convex problems and sub-linear convergence rates for smooth non-convex problems. We conduct numerical experiments demonstrating the robust empirical performance of PerMFL, outperforming the state-of-the-art in multiple personalized federated learning tasks.
Authors: Nihal Acharya Adde, Hanno Gottschalk, Andreas Ebert
Abstract: This paper focuses on hyperparameter optimization for autonomous driving strategies based on Reinforcement Learning. We provide a detailed description of training the RL agent in a simulation environment. Subsequently, we employ Efficient Global Optimization algorithm that uses Gaussian Process fitting for hyperparameter optimization in RL. Before this optimization phase, Gaussian process interpolation is applied to fit the surrogate model, for which the hyperparameter set is generated using Latin hypercube sampling. To accelerate the evaluation, parallelization techniques are employed. Following the hyperparameter optimization procedure, a set of hyperparameters is identified, resulting in a noteworthy enhancement in overall driving performance. There is a substantial increase of 4\% when compared to existing manually tuned parameters and the hyperparameters discovered during the initialization process using Latin hypercube sampling. After the optimization, we analyze the obtained results thoroughly and conduct a sensitivity analysis to assess the robustness and generalization capabilities of the learned autonomous driving strategies. The findings from this study contribute to the advancement of Gaussian process based Bayesian optimization to optimize the hyperparameters for autonomous driving in RL, providing valuable insights for the development of efficient and reliable autonomous driving systems.
Authors: Th\'eo Gnassounou, Antoine Collas, R\'emi Flamary, Karim Lounici, Alexandre Gramfort
Abstract: Machine learning applications on signals such as computer vision or biomedical data often face significant challenges due to the variability that exists across hardware devices or session recordings. This variability poses a Domain Adaptation (DA) problem, as training and testing data distributions often differ. In this work, we propose Spatio-Temporal Monge Alignment (STMA) to mitigate these variabilities. This Optimal Transport (OT) based method adapts the cross-power spectrum density (cross-PSD) of multivariate signals by mapping them to the Wasserstein barycenter of source domains (multi-source DA). Predictions for new domains can be done with a filtering without the need for retraining a model with source data (test-time DA). We also study and discuss two special cases of the method, Temporal Monge Alignment (TMA) and Spatial Monge Alignment (SMA). Non-asymptotic concentration bounds are derived for the mappings estimation, which reveals a bias-plus-variance error structure with a variance decay rate of $\mathcal{O}(n_\ell^{-1/2})$ with $n_\ell$ the signal length. This theoretical guarantee demonstrates the efficiency of the proposed computational schema. Numerical experiments on multivariate biosignals and image data show that STMA leads to significant and consistent performance gains between datasets acquired with very different settings. Notably, STMA is a pre-processing step complementary to state-of-the-art deep learning methods.
Authors: Bart{\l}omiej Krzepkowski, Monika Michaluk, Franciszek Szarwacki, Piotr Kubaty, Jary Pomponi, Tomasz Trzci\'nski, Bartosz W\'ojcik, Kamil Adamczewski
Abstract: Early exits are an important efficiency mechanism integrated into deep neural networks that allows for the termination of the network's forward pass before processing through all its layers. By allowing early halting of the inference process for less complex inputs that reached high confidence, early exits significantly reduce the amount of computation required. Early exit methods add trainable internal classifiers which leads to more intricacy in the training process. However, there is no consistent verification of the approaches of training of early exit methods, and no unified scheme of training such models. Most early exit methods employ a training strategy that either simultaneously trains the backbone network and the exit heads or trains the exit heads separately. We propose a training approach where the backbone is initially trained on its own, followed by a phase where both the backbone and the exit heads are trained together. Thus, we advocate for organizing early-exit training strategies into three distinct categories, and then validate them for their performance and efficiency. In this benchmark, we perform both theoretical and empirical analysis of early-exit training regimes. We study the methods in terms of information flow, loss landscape and numerical rank of activations and gauge the suitability of regimes for various architectures and datasets.
Authors: Mohd Mujtaba Akhtar, Girish, Muskaan Singh, Orchid Chetia Phukan
Abstract: Autism Spectrum Disorder (ASD) is a complex neuro-developmental challenge, presenting a spectrum of difficulties in social interaction, communication, and the expression of repetitive behaviors in different situations. This increasing prevalence underscores the importance of ASD as a major public health concern and the need for comprehensive research initiatives to advance our understanding of the disorder and its early detection methods. This study introduces a novel hierarchical feature fusion method aimed at enhancing the early detection of ASD in children through the analysis of code-switched speech (English and Hindi). Employing advanced audio processing techniques, the research integrates acoustic, paralinguistic, and linguistic information using Transformer Encoders. This innovative fusion strategy is designed to improve classification robustness and accuracy, crucial for early and precise ASD identification. The methodology involves collecting a code-switched speech corpus, CoSAm, from children diagnosed with ASD and a matched control group. The dataset comprises 61 voice recordings from 30 children diagnosed with ASD and 31 from neurotypical children, aged between 3 and 13 years, resulting in a total of 159.75 minutes of voice recordings. The feature analysis focuses on MFCCs and extensive statistical attributes to capture speech pattern variability and complexity. The best model performance is achieved using a hierarchical fusion technique with an accuracy of 98.75% using a combination of acoustic and linguistic features first, followed by paralinguistic features in a hierarchical manner.
Authors: Aidan J. Hughes, Giulia Delo, Jack Poole, Nikolaos Dervilis, Keith Worden
Abstract: In traditional approaches to structural health monitoring, challenges often arise associated with the availability of labelled data. Population-based structural health monitoring seeks to overcomes these challenges by leveraging data/information from similar structures via technologies such as transfer learning. The current paper demonstrate a methodology for quantifying the value of information transfer in the context of operation and maintenance decision-making. This demonstration, based on a population of laboratory-scale aircraft models, highlights the steps required to evaluate the expected value of information transfer including similarity assessment and prediction of transfer efficacy. Once evaluated for a given population, the value of information transfer can be used to optimise transfer-learning strategies for newly-acquired target domains.
Authors: Jiaqi Luo, Yuan Yuan, Shixin Xu
Abstract: Class imbalance remains a significant challenge in machine learning, particularly for tabular data classification tasks. While Gradient Boosting Decision Trees (GBDT) models have proven highly effective for such tasks, their performance can be compromised when dealing with imbalanced datasets. This paper presents the first comprehensive study on adapting class-balanced loss functions to three GBDT algorithms across various tabular classification tasks, including binary, multi-class, and multi-label classification. We conduct extensive experiments on multiple datasets to evaluate the impact of class-balanced losses on different GBDT models, establishing a valuable benchmark. Our results demonstrate the potential of class-balanced loss functions to enhance GBDT performance on imbalanced datasets, offering a robust approach for practitioners facing class imbalance challenges in real-world applications. Additionally, we introduce a Python package that facilitates the integration of class-balanced loss functions into GBDT workflows, making these advanced techniques accessible to a wider audience.
Authors: Rui Xie
Abstract: Deep learning continues to re-shape numerous fields, from natural language processing and imaging to data analytics and recommendation systems. This report studies two research papers that represent recent progress on deep learning from two largely different aspects: The first paper applied the transformer networks, which are typically used in language models, to improve the quality of synthetic aperture radar image by effectively reducing the speckle noise. The second paper presents an in-storage computing design solution to enable cost-efficient and high-performance implementations of deep learning recommendation systems. In addition to summarizing each paper in terms of motivation, key ideas and techniques, and evaluation results, this report also presents thoughts and discussions about possible future research directions. By carrying out in-depth study on these two representative papers and related references, this doctoral candidate has developed better understanding on the far-reaching impact and efficient implementation of deep learning models.
Authors: Aurelio Sulser, Johann Wenckstern, Clara Kuempel
Abstract: We propose GLAudio: Graph Learning on Audio representation of the node features and the connectivity structure. This novel architecture propagates the node features through the graph network according to the discrete wave equation and then employs a sequence learning architecture to learn the target node function from the audio wave signal. This leads to a new paradigm of learning on graph-structured data, in which information propagation and information processing are separated into two distinct steps. We theoretically characterize the expressivity of our model, introducing the notion of the receptive field of a vertex, and investigate our model's susceptibility to over-smoothing and over-squashing both theoretically as well as experimentally on various graph datasets.
Authors: Juliette Decugis, Alicia Y. Tsai, Max Emerling, Ashwin Ganesh, Laurent El Ghaoui
Abstract: In this paper, we investigate the extrapolation capabilities of implicit deep learning models in handling unobserved data, where traditional deep neural networks may falter. Implicit models, distinguished by their adaptability in layer depth and incorporation of feedback within their computational graph, are put to the test across various extrapolation scenarios: out-of-distribution, geographical, and temporal shifts. Our experiments consistently demonstrate significant performance advantage with implicit models. Unlike their non-implicit counterparts, which often rely on meticulous architectural design for each task, implicit models demonstrate the ability to learn complex model structures without the need for task-specific design, highlighting their robustness in handling unseen data.
Authors: Senthooran Rajamanoharan, Tom Lieberum, Nicolas Sonnerat, Arthur Conmy, Vikrant Varma, J\'anos Kram\'ar, Neel Nanda
Abstract: Sparse autoencoders (SAEs) are a promising unsupervised approach for identifying causally relevant and interpretable linear features in a language model's (LM) activations. To be useful for downstream tasks, SAEs need to decompose LM activations faithfully; yet to be interpretable the decomposition must be sparse -- two objectives that are in tension. In this paper, we introduce JumpReLU SAEs, which achieve state-of-the-art reconstruction fidelity at a given sparsity level on Gemma 2 9B activations, compared to other recent advances such as Gated and TopK SAEs. We also show that this improvement does not come at the cost of interpretability through manual and automated interpretability studies. JumpReLU SAEs are a simple modification of vanilla (ReLU) SAEs -- where we replace the ReLU with a discontinuous JumpReLU activation function -- and are similarly efficient to train and run. By utilising straight-through-estimators (STEs) in a principled manner, we show how it is possible to train JumpReLU SAEs effectively despite the discontinuous JumpReLU function introduced in the SAE's forward pass. Similarly, we use STEs to directly train L0 to be sparse, instead of training on proxies such as L1, avoiding problems like shrinkage.
Authors: Jiahong Ma, Mingguo He, Zhewei Wei
Abstract: Spectral Graph Neural Networks have demonstrated superior performance in graph representation learning. However, many current methods focus on employing shared polynomial coefficients for all nodes, i.e., learning node-unified filters, which limits the filters' flexibility for node-level tasks. The recent DSF attempts to overcome this limitation by learning node-wise coefficients based on positional encoding. However, the initialization and updating process of the positional encoding are burdensome, hindering scalability on large-scale graphs. In this work, we propose a scalable node-wise filter, PolyAttn. Leveraging the attention mechanism, PolyAttn can directly learn node-wise filters in an efficient manner, offering powerful representation capabilities. Building on PolyAttn, we introduce the whole model, named PolyFormer. In the lens of Graph Transformer models, PolyFormer, which calculates attention scores within nodes, shows great scalability. Moreover, the model captures spectral information, enhancing expressiveness while maintaining efficiency. With these advantages, PolyFormer offers a desirable balance between scalability and expressiveness for node-level tasks. Extensive experiments demonstrate that our proposed methods excel at learning arbitrary node-wise filters, showing superior performance on both homophilic and heterophilic graphs, and handling graphs containing up to 100 million nodes. The code is available at https://github.com/air029/PolyFormer.
Authors: Xiaotong Sun, Peijie Qiu, Shengfan Zhang
Abstract: Survival analysis models time-to-event distributions with censorship. Recently, deep survival models using neural networks have dominated due to their representational power and state-of-the-art performance. However, their "black-box" nature hinders interpretability, which is crucial in real-world applications. In contrast, "white-box" tree-based survival models offer better interpretability but struggle to converge to global optima due to greedy expansion. In this paper, we bridge the gap between previous deep survival models and traditional tree-based survival models through deep rectified linear unit (ReLU) networks. We show that a deliberately constructed deep ReLU network (SurvReLU) can harness the interpretability of tree-based structures with the representational power of deep survival models. Empirical studies on both simulated and real survival benchmark datasets show the effectiveness of the proposed SurvReLU in terms of performance and interoperability. The code is available at \href{https://github.com/xs018/SurvReLU}{\color{magenta}{ https://github.com/xs018/SurvReLU}}.
URLs: https://github.com/xs018/SurvReLU, https://github.com/xs018/SurvReLU
Authors: Hoang Anh Just, Ming Jin, Anit Sahu, Huy Phan, Ruoxi Jia
Abstract: Reinforcement learning from human feedback plays a crucial role in aligning language models towards human preferences, traditionally represented through comparisons between pairs or sets of responses within a given context. While many studies have enhanced algorithmic techniques to optimize learning from such data, this work shifts focus to improving preference learning through a data-centric approach. Specifically, we propose enriching existing preference datasets with machine-generated rationales that explain the reasons behind choices. We develop a simple and principled framework to augment current preference learning methods with rationale information. Our comprehensive analysis highlights how rationales enhance learning efficiency. Extensive experiments reveal that rationale-enriched preference learning offers multiple advantages: it improves data efficiency, accelerates convergence to higher-performing models, and reduces verbosity bias and hallucination. Furthermore, this framework is versatile enough to integrate with various preference optimization algorithms. Overall, our findings highlight the potential of re-imagining data design for preference learning, demonstrating that even freely available machine-generated rationales can significantly boost performance across multiple dimensions. The code repository is available at https: //github.com/reds-lab/preference-learning-with-rationales
Authors: Rohan Gupta, Iv\'an Arcuschin, Thomas Kwa, Adri\`a Garriga-Alonso
Abstract: Mechanistic interpretability methods aim to identify the algorithm a neural network implements, but it is difficult to validate such methods when the true algorithm is unknown. This work presents InterpBench, a collection of semi-synthetic yet realistic transformers with known circuits for evaluating these techniques. We train these neural networks using a stricter version of Interchange Intervention Training (IIT) which we call Strict IIT (SIIT). Like the original, SIIT trains neural networks by aligning their internal computation with a desired high-level causal model, but it also prevents non-circuit nodes from affecting the model's output. We evaluate SIIT on sparse transformers produced by the Tracr tool and find that SIIT models maintain Tracr's original circuit while being more realistic. SIIT can also train transformers with larger circuits, like Indirect Object Identification (IOI). Finally, we use our benchmark to evaluate existing circuit discovery techniques.
Authors: Rui Luo, Zhixin Zhou
Abstract: This paper introduces Conformal Thresholded Intervals (CTI), a novel conformal regression method that aims to produce the smallest possible prediction set with guaranteed coverage. Unlike existing methods that rely on nested conformal framework and full conditional distribution estimation, CTI estimates the conditional probability density for a new response to fall into each interquantile interval using off-the-shelf multi-output quantile regression. CTI constructs prediction sets by thresholding the estimated conditional interquantile intervals based on their length, which is inversely proportional to the estimated probability density. The threshold is determined using a calibration set to ensure marginal coverage. Experimental results demonstrate that CTI achieves optimal performance across various datasets.
Authors: Prasenjit Karmakar, Swadhin Pradhan, Sandip Chakraborty
Abstract: In recent years, indoor air pollution has posed a significant threat to our society, claiming over 3.2 million lives annually. Developing nations, such as India, are most affected since lack of knowledge, inadequate regulation, and outdoor air pollution lead to severe daily exposure to pollutants. However, only a limited number of studies have attempted to understand how indoor air pollution affects developing countries like India. To address this gap, we present spatiotemporal measurements of air quality from 30 indoor sites over six months during summer and winter seasons. The sites are geographically located across four regions of type: rural, suburban, and urban, covering the typical low to middle-income population in India. The dataset contains various types of indoor environments (e.g., studio apartments, classrooms, research laboratories, food canteens, and residential households), and can provide the basis for data-driven learning model research aimed at coping with unique pollution patterns in developing countries. This unique dataset demands advanced data cleaning and imputation techniques for handling missing data due to power failure or network outages during data collection. Furthermore, through a simple speech-to-text application, we provide real-time indoor activity labels annotated by occupants. Therefore, environmentalists and ML enthusiasts can utilize this dataset to understand the complex patterns of the pollutants under different indoor activities, identify recurring sources of pollution, forecast exposure, improve floor plans and room structures of modern indoor designs, develop pollution-aware recommender systems, etc.
Authors: Thomas Kwa, Drake Thomas, Adri\`a Garriga-Alonso
Abstract: When applying reinforcement learning from human feedback (RLHF), the reward is learned from data and, therefore, always has some error. It is common to mitigate this by regularizing the policy with KL divergence from a base model, with the hope that balancing reward with regularization will achieve desirable outcomes despite this reward misspecification. We show that when the reward function has light-tailed error, optimal policies under less restrictive KL penalties achieve arbitrarily high utility. However, if error is heavy-tailed, some policies obtain arbitrarily high reward despite achieving no more utility than the base model--a phenomenon we call catastrophic Goodhart. We adapt a discrete optimization method to measure the tails of reward models, finding that they are consistent with light-tailed error. However, the pervasiveness of heavy-tailed distributions in many real-world applications indicates that future sources of RL reward could have heavy-tailed error, increasing the likelihood of reward hacking even with KL regularization.
Authors: Yiming Zhou, Callen MacPhee, Tingyi Zhou, Bahram Jalali
Abstract: Deep neural networks (DNNs) have achieved exceptional performance across various fields by learning complex nonlinear mappings from large-scale datasets. However, they encounter challenges such as high computational costs and limited interpretability. To address these issues, hybrid approaches that integrate physics with AI are gaining interest. This paper introduces a novel physics-based AI model called the "Nonlinear Schr\"odinger Network", which treats the Nonlinear Schr\"odinger Equation (NLSE) as a general-purpose trainable model for learning complex patterns including nonlinear mappings and memory effects from data. Existing physics-informed machine learning methods use neural networks to approximate the solutions of partial differential equations (PDEs). In contrast, our approach directly treats the PDE as a trainable model to obtain general nonlinear mappings that would otherwise require neural networks. As a physics-inspired approach, it offers a more interpretable and parameter-efficient alternative to traditional black-box neural networks, achieving comparable or better accuracy in time series classification tasks while significantly reducing the number of required parameters. Notably, the trained Nonlinear Schr\"odinger Network is interpretable, with all parameters having physical meanings as properties of a virtual physical system that transforms the data to a more separable space. This interpretability allows for insight into the underlying dynamics of the data transformation process. Applications to time series forecasting have also been explored. While our current implementation utilizes the NLSE, the proposed method of using physics equations as trainable models to learn nonlinear mappings from data is not limited to the NLSE and may be extended to other master equations of physics.
Authors: Adrian Shuai Li, Elisa Bertino, Xuan-Hong Dang, Ankush Singla, Yuhai Tu, Mark N Wegman
Abstract: The most effective domain adaptation (DA) involves the decomposition of data representation into a domain independent representation (DIRep), and a domain dependent representation (DDRep). A classifier is trained by using the DIRep of the labeled source images. Since the DIRep is domain invariant, the classifier can be "transferred" to make predictions for the target domain with no (or few) labels. However, information useful for classification in the target domain can "hide" in the DDRep in current DA algorithms such as Domain-Separation-Networks (DSN). DSN's weak constraint to enforce orthogonality of DIRep and DDRep, allows this hiding and can result in poor performance. To address this shortcoming, we developed a new algorithm wherein a stronger constraint is imposed to minimize the DDRep by using a KL divergent loss for the DDRep in order to create the maximal DIRep that enhances transfer learning performance. By using synthetic data sets, we show explicitly that depending on initialization DSN with its weaker constraint can lead to sub-optimal solutions with poorer DA performance whereas our algorithm with maximal DIRep is robust against such perturbations. We demonstrate the equal-or-better performance of our approach against state-of-the-art algorithms by using several standard benchmark image datasets including Office. We further highlight the compatibility of our algorithm with pretrained models, extending its applicability and versatility in real-world scenarios.
Authors: Ian Hough, Lo\"ic Argentier, Ziyang Jiang, Tongshu Zheng, Mike Bergin, David Carlson, Jean-Luc Jaffrezo, Jocelyn Chanussot, Ga\"elle Uzu
Abstract: Oxidative potential (OP), which measures particulate matter's (PM) capacity to induce oxidative stress in the lungs, is increasingly recognized as an indicator of PM toxicity. Since OP is not routinely monitored, it can be challenging to estimate exposure and health impacts. Remote sensing data are commonly used to estimate PM mass concentration, but have never been used to estimate OP. In this study, we evaluate the potential of satellite images to estimate OP as measured by acellular ascorbic acid (OP AA) and dithiothreitol (OP DTT) assays of 24-hour PM10 sampled periodically over five years at three locations around Grenoble, France. We use a deep convolutional neural network to extract features of daily 3 m/pixel PlanetScope satellite images and train a multilayer perceptron to estimate OP at a 1 km spatial resolution based on the image features and common meteorological variables. The model captures more than half of the variation in OP AA and almost half of the variation in OP DTT (test set R2 = 0.62 and 0.48, respectively), with relative mean absolute error (MAE) of about 32%. Using only satellite images, the model still captures about half of the variation in OP AA and one third of the variation in OP DTT (test set R2 = 0.49 and 0.36, respectively) with relative MAE of about 37%. If confirmed in other areas, our approach could represent a low-cost method for expanding the temporal or spatial coverage of OP estimates.
Authors: Siddartha Reddy N, Sai Prakash MV, Varun V, Vishal Vaddina, Saisubramaniam Gopalakrishnan
Abstract: Lead optimization is a pivotal task in the drug design phase within the drug discovery lifecycle. The primary objective is to refine the lead compound to meet specific molecular properties for progression to the subsequent phase of development. In this work, we present an innovative approach, Latent Evolutionary Optimization for Molecule Generation (LEOMol), a generative modeling framework for the efficient generation of optimized molecules. LEOMol leverages Evolutionary Algorithms, such as Genetic Algorithm and Differential Evolution, to search the latent space of a Variational AutoEncoder (VAE). This search facilitates the identification of the target molecule distribution within the latent space. Our approach consistently demonstrates superior performance compared to previous state-of-the-art models across a range of constrained molecule generation tasks, outperforming existing models in all four sub-tasks related to property targeting. Additionally, we suggest the importance of including toxicity in the evaluation of generative models. Furthermore, an ablation study underscores the improvements that our approach provides over gradient-based latent space optimization methods. This underscores the effectiveness and superiority of LEOMol in addressing the inherent challenges in constrained molecule generation while emphasizing its potential to propel advancements in drug discovery.
Authors: Ulrich A. Mbou Sob, Qiulin Li, Miguel Arbes\'u, Oliver Bent, Andries P. Smit, Arnu Pretorius
Abstract: A specific challenge with deep learning approaches for molecule generation is generating both syntactically valid and chemically plausible molecular string representations. To address this, we propose a novel generative latent-variable transformer model for small molecules that leverages a recently proposed molecular string representation called SAFE. We introduce a modification to SAFE to reduce the number of invalid fragmented molecules generated during training and use this to train our model. Our experiments show that our model can generate novel molecules with a validity rate > 90% and a fragmentation rate < 1% by sampling from a latent space. By fine-tuning the model using reinforcement learning to improve molecular docking, we significantly increase the number of hit candidates for five specific protein targets compared to the pre-trained model, nearly doubling this number for certain targets. Additionally, our top 5% mean docking scores are comparable to the current state-of-the-art (SOTA), and we marginally outperform SOTA on three of the five targets.
Authors: Ali Ghiasvand Mohammadkhani
Abstract: Recently, various encoder-only and encoder-decoder pre-trained models like BERT and T5 have been applied to automatic essay scoring (AES) as small language models. However, existing studies have primarily treated this task akin to a classification problem, focusing solely on outputting scores in the target text without offering interpretations for the generated scores. Departing from the approaches, we introduce Reasoning Distillation-Based Evaluation (RDBE), which integrates interpretability to elucidate the rationale behind model scores while enhancing performance through initial reasoning. This interpretive capability is acquired during training by leveraging generated reasoning from a large language model (LLM) to distill a small language model (SLM). Our experimental results demonstrate the efficacy of RDBE across all scoring rubrics considered in the dataset. RDBE outperforms both zero-shot LLM generation and generation from a baseline fine-tuned model, establishing itself as state-of-the-art in the corresponding dataset. This highlights its practical interpretative output and enhanced performance.
Authors: Blake A. Wilson, Jonathan Wurtz, Vahagn Mkhitaryan, Michael Bezick, Sheng-Tao Wang, Sabre Kais, Vladimir M. Shalaev, Alexandra Boltasseva
Abstract: Large-scale optimization problems are prevalent in several fields, including engineering, finance, and logistics. However, most optimization problems cannot be efficiently encoded onto a physical system because the existing quantum samplers have too few qubits. Another typical limiting factor is that the optimization constraints are not compatible with the native cost Hamiltonian. This work presents a new approach to address these challenges. We introduce the adversarial quantum autoencoder model (AQAM) that can be used to map large-scale optimization problems onto existing quantum samplers while simultaneously optimizing the problem through latent quantum-enhanced Boltzmann sampling. We demonstrate the AQAM on a neutral atom sampler, and showcase the model by optimizing 64px by 64px unit cells that represent a broad-angle filter metasurface applicable to improving the coherence of neutral atom devices. Using 12-atom simulations, we demonstrate that the AQAM achieves a lower Renyi divergence and a larger spectral gap when compared to classical Markov Chain Monte Carlo samplers. Our work paves the way to more efficient mapping of conventional optimization problems into existing quantum samplers.
Authors: Rahul Ramesh, Anthony Bisulco, Ronald W. DiTullio, Linran Wei, Vijay Balasubramanian, Kostas Daniilidis, Pratik Chaudhari
Abstract: We show that many perception tasks, from visual recognition, semantic segmentation, optical flow, depth estimation to vocalization discrimination, are highly redundant functions of their input data. Images or spectrograms, projected into different subspaces, formed by orthogonal bases in pixel, Fourier or wavelet domains, can be used to solve these tasks remarkably well regardless of whether it is the top subspace where data varies the most, some intermediate subspace with moderate variability--or the bottom subspace where data varies the least. This phenomenon occurs because different subspaces have a large degree of redundant information relevant to the task.
Authors: Sirnam Swetha, Jinyu Yang, Tal Neiman, Mamshad Nayeem Rizve, Son Tran, Benjamin Yao, Trishul Chilimbi, Mubarak Shah
Abstract: Recent advancements in Multimodal Large Language Models (MLLMs) have revolutionized the field of vision-language understanding by integrating visual perception capabilities into Large Language Models (LLMs). The prevailing trend in this field involves the utilization of a vision encoder derived from vision-language contrastive learning (CL), showing expertise in capturing overall representations while facing difficulties in capturing detailed local patterns. In this work, we focus on enhancing the visual representations for MLLMs by combining high-frequency and detailed visual representations, obtained through masked image modeling (MIM), with semantically-enriched low-frequency representations captured by CL. To achieve this goal, we introduce X-Former which is a lightweight transformer module designed to exploit the complementary strengths of CL and MIM through an innovative interaction mechanism. Specifically, X-Former first bootstraps vision-language representation learning and multimodal-to-multimodal generative learning from two frozen vision encoders, i.e., CLIP-ViT (CL-based) and MAE-ViT (MIM-based). It further bootstraps vision-to-language generative learning from a frozen LLM to ensure visual features from X-Former can be interpreted by the LLM. To demonstrate the effectiveness of our approach, we assess its performance on tasks demanding detailed visual understanding. Extensive evaluations indicate that X-Former excels in visual reasoning tasks involving both structural and semantic categories in the GQA dataset. Assessment on fine-grained visual perception benchmark further confirms its superior capabilities in visual understanding.
Authors: Mohammad Aamir Sohail, Mohsen Heidari Khoozani, S. Sandeep Pradhan
Abstract: Quantum machine learning through variational quantum algorithms (VQAs) has gained substantial attention in recent years. VQAs employ parameterized quantum circuits, which are typically optimized using gradient-based methods. However, these methods often exhibit sub-optimal convergence performance due to their dependence on Euclidean geometry. The quantum natural gradient descent (QNGD) optimization method, which considers the geometry of the quantum state space via a quantum information (Riemannian) metric tensor, provides a more effective optimization strategy. Despite its advantages, QNGD encounters notable challenges for learning from quantum data, including the no-cloning principle, which prohibits the replication of quantum data, state collapse, and the measurement postulate, which leads to the stochastic loss function. This paper introduces the quantum natural stochastic pairwise coordinate descent (2-QNSCD) optimization method. This method leverages the curved geometry of the quantum state space through a novel ensemble-based quantum information metric tensor, offering a more physically realizable optimization strategy for learning from quantum data. To improve computational efficiency and reduce sample complexity, we develop a highly sparse unbiased estimator of the novel metric tensor using a quantum circuit with gate complexity $\Theta(1)$ times that of the parameterized quantum circuit and single-shot quantum measurements. Our approach avoids the need for multiple copies of quantum data, thus adhering to the no-cloning principle. We provide a detailed theoretical foundation for our optimization method, along with an exponential convergence analysis. Additionally, we validate the utility of our method through a series of numerical experiments.
Authors: Michael J. Bianco, David Eigen, Michael Gormish
Abstract: Geolocating images of a ground-level scene entails estimating the location on Earth where the picture was taken, in absence of GPS or other location metadata. Typically, methods are evaluated by measuring the Great Circle Distance (GCD) between a predicted location and ground truth. However, this measurement is limited because it only evaluates a single point, not estimates of regions or score heatmaps. This is especially important in applications to rural, wilderness and under-sampled areas, where finding the exact location may not be possible, and when used in aggregate systems that progressively narrow down locations. In this paper, we introduce a novel metric, Recall vs Area (RvA), which measures the accuracy of estimated distributions of locations. RvA treats image geolocation results similarly to document retrieval, measuring recall as a function of area: For a ranked list of (possibly non-contiguous) predicted regions, we measure the accumulated area required for the region to contain the ground truth coordinate. This produces a curve similar to a precision-recall curve, where "precision" is replaced by square kilometers area, allowing evaluation of performance for different downstream search area budgets. Following directly from this view of the problem, we then examine a simple ensembling approach to global-scale image geolocation, which incorporates information from multiple sources to help address domain shift, and can readily incorporate multiple models, attribute predictors, and data sources. We study its effectiveness by combining the geolocation models GeoEstimation and the current SOTA GeoCLIP, with attribute predictors based on ORNL LandScan and ESA-CCI Land Cover. We find significant improvements in image geolocation for areas that are under-represented in the training set, particularly non-urban areas, on both Im2GPS3k and Street View images.
Authors: Madhava Krishna, A V Subramanyam
Abstract: SimMIM is a widely used method for pretraining vision transformers using masked image modeling. However, despite its success in fine-tuning performance, it has been shown to perform sub-optimally when used for linear probing. We propose an efficient patch-wise weighting derived from keypoint features which captures the local information and provides better context during SimMIM's reconstruction phase. Our method, KAMIM, improves the top-1 linear probing accuracy from 16.12% to 33.97%, and finetuning accuracy from 76.78% to 77.3% when tested on the ImageNet-1K dataset with a ViT-B when trained for the same number of epochs. We conduct extensive testing on different datasets, keypoint extractors, and model architectures and observe that patch-wise weighting augments linear probing performance for larger pretraining datasets. We also analyze the learned representations of a ViT-B trained using KAMIM and observe that they behave similar to contrastive learning with regard to its behavior, with longer attention distances and homogenous self-attention across layers. Our code is publicly available at https://github.com/madhava20217/KAMIM.
Authors: Sitan Chen, Jerry Li, Allen Liu
Abstract: We give the first tight sample complexity bounds for shadow tomography and classical shadows in the regime where the target error is below some sufficiently small inverse polynomial in the dimension of the Hilbert space. Formally we give a protocol that, given any $m\in\mathbb{N}$ and $\epsilon \le O(d^{-12})$, measures $O(\log(m)/\epsilon^2)$ copies of an unknown mixed state $\rho\in\mathbb{C}^{d\times d}$ and outputs a classical description of $\rho$ which can then be used to estimate any collection of $m$ observables to within additive accuracy $\epsilon$. Previously, even for the simpler task of shadow tomography -- where the $m$ observables are known in advance -- the best known rates either scaled benignly but suboptimally in all of $m, d, \epsilon$, or scaled optimally in $\epsilon, m$ but had additional polynomial factors in $d$ for general observables. Intriguingly, we also show via dimensionality reduction, that we can rescale $\epsilon$ and $d$ to reduce to the regime where $\epsilon \le O(d^{-1/2})$. Our algorithm draws upon representation-theoretic tools recently developed in the context of full state tomography.
Authors: Michael Oliver, Guan Wang
Abstract: This paper addresses the challenges of efficiently fine-tuning large language models (LLMs) by exploring data efficiency and hyperparameter optimization. We investigate the minimum data required for effective fine-tuning and propose a novel hyperparameter optimization method that leverages early-stage model performance. Our experiments demonstrate that fine-tuning with as few as 200 samples can improve model accuracy from 70\% to 88\% in a product attribute extraction task. We identify a saturation point of approximately 6,500 samples, beyond which additional data yields diminishing returns. Our proposed bayesian hyperparameter optimization method, which evaluates models at 20\% of total training time, correlates strongly with final model performance, with 4 out of 5 top early-stage models remaining in the top 5 at completion. This approach led to a 2\% improvement in accuracy over baseline models when evaluated on an independent test set. These findings offer actionable insights for practitioners, potentially reducing computational load and dependency on extensive datasets while enhancing overall performance of fine-tuned LLMs.
Authors: Qifan Zhang, Yunhui Guo, Yu Xiang
Abstract: We study the problem of Continual Distillation Learning (CDL) that considers Knowledge Distillation (KD) in the Continual Learning (CL) setup. A teacher model and a student model need to learn a sequence of tasks, and the knowledge of the teacher model will be distilled to the student to improve the student model. We introduce a novel method named CDL-Prompt that utilizes prompt-based continual learning models to build the teacher-student model. We investigate how to utilize the prompts of the teacher model in the student model for knowledge distillation, and propose an attention-based prompt mapping scheme to use the teacher prompts for the student. We demonstrate that our method can be applied to different prompt-based continual learning models such as L2P, DualPrompt and CODA-Prompt to improve their performance using powerful teacher models. Although recent CL methods focus on prompt learning, we show that our method can be utilized to build efficient CL models using prompt-based knowledge distillation.
Authors: Guruprasad V Ramesh, Harrison Rosenberg, Ashish Hooda, Kassem Fawaz
Abstract: Computer vision systems have been deployed in various applications involving biometrics like human faces. These systems can identify social media users, search for missing persons, and verify identity of individuals. While computer vision models are often evaluated for accuracy on available benchmarks, more annotated data is necessary to learn about their robustness and fairness against semantic distributional shifts in input data, especially in face data. Among annotated data, counterfactual examples grant strong explainability characteristics. Because collecting natural face data is prohibitively expensive, we put forth a generative AI-based framework to construct targeted, counterfactual, high-quality synthetic face data. Our synthetic data pipeline has many use cases, including face recognition systems sensitivity evaluations and image understanding system probes. The pipeline is validated with multiple user studies. We showcase the efficacy of our face generation pipeline on a leading commercial vision model. We identify facial attributes that cause vision systems to fail.
Authors: Paolo Calafiura, Jay Chan, Loic Delabrouille, Brandon Wang
Abstract: Track reconstruction is a crucial task in particle experiments and is traditionally very computationally expensive due to its combinatorial nature. Recently, graph neural networks (GNNs) have emerged as a promising approach that can improve scalability. Most of these GNN-based methods, including the edge classification (EC) and the object condensation (OC) approach, require an input graph that needs to be constructed beforehand. In this work, we consider a one-shot OC approach that reconstructs particle tracks directly from a set of hits (point cloud) by recursively applying graph attention networks with an evolving graph structure. This approach iteratively updates the graphs and can better facilitate the message passing across each graph. Preliminary studies on the TrackML dataset show better track performance compared to the methods that require a fixed input graph.
Authors: James Giroux, Ariyarathne Gangani, Alexander C. Nwala, Cristiano Fanelli
Abstract: Social bots remain a major vector for spreading disinformation on social media and a menace to the public. Despite the progress made in developing multiple sophisticated social bot detection algorithms and tools, bot detection remains a challenging, unsolved problem that is fraught with uncertainty due to the heterogeneity of bot behaviors, training data, and detection algorithms. Detection models often disagree on whether to label the same account as bot or human-controlled. However, they do not provide any measure of uncertainty to indicate how much we should trust their results. We propose to address both bot detection and the quantification of uncertainty at the account level - a novel feature of this research. This dual focus is crucial as it allows us to leverage additional information related to the quantified uncertainty of each prediction, thereby enhancing decision-making and improving the reliability of bot classifications. Specifically, our approach facilitates targeted interventions for bots when predictions are made with high confidence and suggests caution (e.g., gathering more data) when predictions are uncertain.
Authors: Junghyun Lee, Se-Young Yun, Kwang-Sung Jun
Abstract: We present a unified likelihood ratio-based confidence sequence (CS) for any (self-concordant) generalized linear models (GLMs) that is guaranteed to be convex and numerically tight. We show that this is on par or improves upon known CSs for various GLMs, including Gaussian, Bernoulli, and Poisson. In particular, for the first time, our CS for Bernoulli has a poly(S)-free radius where S is the norm of the unknown parameter. Our first technical novelty is its derivation, which utilizes a time-uniform PAC-Bayesian bound with a uniform prior/posterior, despite the latter being a rather unpopular choice for deriving CSs. As a direct application of our new CS, we propose a simple and natural optimistic algorithm called OFUGLB applicable to any generalized linear bandits (GLB; Filippi et al. (2010)). Our analysis shows that the celebrated optimistic approach simultaneously attains state-of-the-art regrets for various self-concordant (not necessarily bounded) GLBs, and even poly(S)-free for bounded GLBs, including logistic bandits. The regret analysis, our second technical novelty, follows from combining our new CS with a new proof technique that completely avoids the previously widely used self-concordant control lemma (Faury et al., 2020, Lemma 9). Finally, we verify numerically that OFUGLB significantly outperforms the prior state-of-the-art (Lee et al., 2024) for logistic bandits.
Authors: Qiong Zhang, Jiahua Chen
Abstract: This paper proposes two split-and-conquer (SC) learning estimators for finite mixture models that are tolerant to Byzantine failures. In SC learning, individual machines obtain local estimates, which are then transmitted to a central server for aggregation. During this communication, the server may receive malicious or incorrect information from some local machines, a scenario known as Byzantine failures. While SC learning approaches have been devised to mitigate Byzantine failures in statistical models with Euclidean parameters, developing Byzantine-tolerant methods for finite mixture models with non-Euclidean parameters requires a distinct strategy. Our proposed distance-based methods are hyperparameter tuning free, unlike existing methods, and are resilient to Byzantine failures while achieving high statistical efficiency. We validate the effectiveness of our methods both theoretically and empirically via experiments on simulated and real data from machine learning applications for digit recognition. The code for the experiment can be found at https://github.com/SarahQiong/RobustSCGMM.
Authors: Xiwei Cheng, Xiangxin Zhou, Yuwei Yang, Yu Bao, Quanquan Gu
Abstract: Diffusion models have achieved promising results for Structure-Based Drug Design (SBDD). Nevertheless, high-quality protein subpocket and ligand data are relatively scarce, which hinders the models' generation capabilities. Recently, Direct Preference Optimization (DPO) has emerged as a pivotal tool for the alignment of generative models such as large language models and diffusion models, providing greater flexibility and accuracy by directly aligning model outputs with human preferences. Building on this advancement, we introduce DPO to SBDD in this paper. We tailor diffusion models to pharmaceutical needs by aligning them with elaborately designed chemical score functions. We propose a new structure-based molecular optimization method called DecompDPO, which decomposes the molecule into arms and scaffolds and performs preference optimization at both local substructure and global molecule levels, allowing for more precise control with fine-grained preferences. Notably, DecompDPO can be effectively used for two main purposes: (1) fine-tuning pretrained diffusion models for molecule generation across various protein families, and (2) molecular optimization given a specific protein subpocket after generation. Extensive experiments on the CrossDocked2020 benchmark show that DecompDPO significantly improves model performance in both molecule generation and optimization, with up to 100% Median High Affinity and a 54.9% Success Rate.
Authors: Md Sultan Al Nahian, Ramakanth Kavuluru
Abstract: Extractive question answering over clinical text is a crucial need to help deal with the deluge of clinical text generated in hospitals. While encoder models (e.g., BERT) have been popular for this reading comprehension task, recently encoder-decoder models (e.g., T5) are on the rise. There is also the emergence of preference optimization techniques to align decoder-only LLMs with human preferences. In this paper, we combine encoder-decoder models with the direct preference optimization (DPO) method to improve over prior state of the art for the RadQA radiology question answering task by 12-15 F1 points. To the best of our knowledge, this effort is the first to show that DPO method also works for reading comprehension via novel heuristics to generate preference data without human inputs.
Authors: Zhenghao Li, Sanyou Wu, Long Feng
Abstract: This paper focuses on the analysis of sequential image data, particularly brain imaging data such as MRI, fMRI, CT, with the motivation of understanding the brain aging process and neurodegenerative diseases. To achieve this goal, we investigate image generation in a time series context. Specifically, we formulate a min-max problem derived from the $f$-divergence between neighboring pairs to learn a time series generator in a nonparametric manner. The generator enables us to generate future images by transforming prior lag-k observations and a random vector from a reference distribution. With a deep neural network learned generator, we prove that the joint distribution of the generated sequence converges to the latent truth under a Markov and a conditional invariance condition. Furthermore, we extend our generation mechanism to a panel data scenario to accommodate multiple samples. The effectiveness of our mechanism is evaluated by generating real brain MRI sequences from the Alzheimer's Disease Neuroimaging Initiative. These generated image sequences can be used as data augmentation to enhance the performance of further downstream tasks, such as Alzheimer's disease detection.
Authors: Fengyu Yang, Chao Feng, Daniel Wang, Tianye Wang, Ziyao Zeng, Zhiyang Xu, Hyoungseob Park, Pengliang Ji, Hanbin Zhao, Yuanning Li, Alex Wong
Abstract: Understanding neural activity and information representation is crucial for advancing knowledge of brain function and cognition. Neural activity, measured through techniques like electrophysiology and neuroimaging, reflects various aspects of information processing. Recent advances in deep neural networks offer new approaches to analyzing these signals using pre-trained models. However, challenges arise due to discrepancies between different neural signal modalities and the limited scale of high-quality neural data. To address these challenges, we present NeuroBind, a general representation that unifies multiple brain signal types, including EEG, fMRI, calcium imaging, and spiking data. To achieve this, we align neural signals in these image-paired neural datasets to pre-trained vision-language embeddings. Neurobind is the first model that studies different neural modalities interconnectedly and is able to leverage high-resource modality models for various neuroscience tasks. We also showed that by combining information from different neural signal modalities, NeuroBind enhances downstream performance, demonstrating the effectiveness of the complementary strengths of different neural modalities. As a result, we can leverage multiple types of neural signals mapped to the same space to improve downstream tasks, and demonstrate the complementary strengths of different neural modalities. This approach holds significant potential for advancing neuroscience research, improving AI systems, and developing neuroprosthetics and brain-computer interfaces.
Authors: Yingrong Wang, Haoxuan Li, Minqin Zhu, Anpeng Wu, Ruoxuan Xiong, Fei Wu, Kun Kuang
Abstract: Causal inference plays an important role in explanatory analysis and decision making across various fields like statistics, marketing, health care, and education. Its main task is to estimate treatment effects and make intervention policies. Traditionally, most of the previous works typically focus on the binary treatment setting that there is only one treatment for a unit to adopt or not. However, in practice, the treatment can be much more complex, encompassing multi-valued, continuous, or bundle options. In this paper, we refer to these as complex treatments and systematically and comprehensively review the causal inference methods for addressing them. First, we formally revisit the problem definition, the basic assumptions, and their possible variations under specific conditions. Second, we sequentially review the related methods for multi-valued, continuous, and bundled treatment settings. In each situation, we tentatively divide the methods into two categories: those conforming to the unconfoundedness assumption and those violating it. Subsequently, we discuss the available datasets and open-source codes. Finally, we provide a brief summary of these works and suggest potential directions for future research.
Authors: Fei Zhu, Xu-Yao Zhang, Zhen Cheng, Cheng-Lin Liu
Abstract: Class-incremental learning (CIL) aims to recognize new classes incrementally while maintaining the discriminability of old classes. Most existing CIL methods are exemplar-based, i.e., storing a part of old data for retraining. Without relearning old data, those methods suffer from catastrophic forgetting. In this paper, we figure out two inherent problems in CIL, i.e., representation bias and classifier bias, that cause catastrophic forgetting of old knowledge. To address these two biases, we present a simple and novel dual bias reduction framework that employs self-supervised transformation (SST) in input space and prototype augmentation (protoAug) in deep feature space. On the one hand, SST alleviates the representation bias by learning generic and diverse representations that can transfer across different tasks. On the other hand, protoAug overcomes the classifier bias by explicitly or implicitly augmenting prototypes of old classes in the deep feature space, which poses tighter constraints to maintain previously learned decision boundaries. We further propose hardness-aware prototype augmentation and multi-view ensemble strategies, leading to significant improvements. The proposed framework can be easily integrated with pre-trained models. Without storing any samples of old classes, our method can perform comparably with state-of-the-art exemplar-based approaches which store plenty of old data. We hope to draw the attention of researchers back to non-exemplar CIL by rethinking the necessity of storing old samples in CIL.
Authors: Prerana Sanjay Kulkarni, Muskaan Jain, Disha Sheshanarayana, Srinivasan Parthiban
Abstract: Despite advancements in drug development strategies, 90% of clinical trials fail. This suggests overlooked aspects in target validation and drug optimization. In order to address this, we introduce HeCiX-KG, Hetionet-Clinicaltrials neXus Knowledge Graph, a novel fusion of data from ClinicalTrials.gov and Hetionet in a single knowledge graph. HeCiX-KG combines data on previously conducted clinical trials from ClinicalTrials.gov, and domain expertise on diseases and genes from Hetionet. This offers a thorough resource for clinical researchers. Further, we introduce HeCiX, a system that uses LangChain to integrate HeCiX-KG with GPT-4, and increase its usability. HeCiX shows high performance during evaluation against a range of clinically relevant issues, proving this model to be promising for enhancing the effectiveness of clinical research. Thus, this approach provides a more holistic view of clinical trials and existing biological data.
Authors: Pradyumna Saligram, Andrew Lanpouthakoun
Abstract: We explore advanced fine-tuning techniques to boost BERT's performance in sentiment analysis, paraphrase detection, and semantic textual similarity. Our approach leverages SMART regularization to combat overfitting, improves hyperparameter choices, employs a cross-embedding Siamese architecture for improved sentence embeddings, and introduces innovative early exiting methods. Our fine-tuning findings currently reveal substantial improvements in model efficiency and effectiveness when combining multiple fine-tuning architectures, achieving a state-of-the-art performance score of on the test set, surpassing current benchmarks and highlighting BERT's adaptability in multifaceted linguistic tasks.
Authors: Peiyong Wang, Casey R. Myers, Lloyd C. L. Hollenberg, Udaya Parampalli
Abstract: When applying quantum computing to machine learning tasks, one of the first considerations is the design of the quantum machine learning model itself. Conventionally, the design of quantum machine learning algorithms relies on the ``quantisation" of classical learning algorithms, such as using quantum linear algebra to implement important subroutines of classical algorithms, if not the entire algorithm, seeking to achieve quantum advantage through possible run-time accelerations brought by quantum computing. However, recent research has started questioning whether quantum advantage via speedup is the right goal for quantum machine learning [1]. Research also has been undertaken to exploit properties that are unique to quantum systems, such as quantum contextuality, to better design quantum machine learning models [2]. In this paper, we take an alternative approach by incorporating the heuristics and empirical evidences from the design of classical deep learning algorithms to the design of quantum neural networks. We first construct a model based on the data reuploading circuit [3] with the quantum Hamiltonian data embedding unitary [4]. Through numerical experiments on images datasets, including the famous MNIST and FashionMNIST datasets, we demonstrate that our model outperforms the quantum convolutional neural network (QCNN)[5] by a large margin (up to over 40% on MNIST test set). Based on the model design process and numerical results, we then laid out six principles for designing quantum machine learning models, especially quantum neural networks.
Authors: Praveen Srinivasa Varadhan, Ashwin Sankar, Giri Raju, Mitesh M. Khapra
Abstract: We release Rasa, the first multilingual expressive TTS dataset for any Indian language, which contains 10 hours of neutral speech and 1-3 hours of expressive speech for each of the 6 Ekman emotions covering 3 languages: Assamese, Bengali, & Tamil. Our ablation studies reveal that just 1 hour of neutral and 30 minutes of expressive data can yield a Fair system as indicated by MUSHRA scores. Increasing neutral data to 10 hours, with minimal expressive data, significantly enhances expressiveness. This offers a practical recipe for resource-constrained languages, prioritizing easily obtainable neutral data alongside smaller amounts of expressive data. We show the importance of syllabically balanced data and pooling emotions to enhance expressiveness. We also highlight challenges in generating specific emotions, e.g., fear and surprise.
Authors: Qichen Fu, Minsik Cho, Thomas Merth, Sachin Mehta, Mohammad Rastegari, Mahyar Najibi
Abstract: The inference of transformer-based large language models consists of two sequential stages: 1) a prefilling stage to compute the KV cache of prompts and generate the first token, and 2) a decoding stage to generate subsequent tokens. For long prompts, the KV cache must be computed for all tokens during the prefilling stage, which can significantly increase the time needed to generate the first token. Consequently, the prefilling stage may become a bottleneck in the generation process. An open question remains whether all prompt tokens are essential for generating the first token. To answer this, we introduce a novel method, LazyLLM, that selectively computes the KV for tokens important for the next token prediction in both the prefilling and decoding stages. Contrary to static pruning approaches that prune the prompt at once, LazyLLM allows language models to dynamically select different subsets of tokens from the context in different generation steps, even though they might be pruned in previous steps. Extensive experiments on standard datasets across various tasks demonstrate that LazyLLM is a generic method that can be seamlessly integrated with existing language models to significantly accelerate the generation without fine-tuning. For instance, in the multi-document question-answering task, LazyLLM accelerates the prefilling stage of the LLama 2 7B model by 2.34x while maintaining accuracy.
Authors: Wenxuan Lu, Mengshun Hu, Yansheng Qiu, Liang Liao, Zheng Wang
Abstract: With the development of VR-related techniques, viewers can enjoy a realistic and immersive experience through a head-mounted display, while omnidirectional video with a low frame rate can lead to user dizziness. However, the prevailing plane frame interpolation methodologies are unsuitable for Omnidirectional Video Interpolation, chiefly due to the lack of models tailored to such videos with strong distortion, compounded by the scarcity of valuable datasets for Omnidirectional Video Frame Interpolation. In this paper, we introduce the benchmark dataset, 360VFI, for Omnidirectional Video Frame Interpolation. We present a practical implementation that introduces a distortion prior from omnidirectional video into the network to modulate distortions. We especially propose a pyramid distortion-sensitive feature extractor that uses the unique characteristics of equirectangular projection (ERP) format as prior information. Moreover, we devise a decoder that uses an affine transformation to facilitate the synthesis of intermediate frames further. 360VFI is the first dataset and benchmark that explores the challenge of Omnidirectional Video Frame Interpolation. Through our benchmark analysis, we presented four different distortion conditions scenes in the proposed 360VFI dataset to evaluate the challenge triggered by distortion during interpolation. Besides, experimental results demonstrate that Omnidirectional Video Interpolation can be effectively improved by modeling for omnidirectional distortion.
Authors: Tao Lin, Kun Jin, Andrew Estornell, Xiaoying Zhang, Yiling Chen, Yang Liu
Abstract: Recommender systems present relevant contents to users and help content creators reach their target audience. The dual nature of these systems influences both users and creators: users' preferences are affected by the items they are recommended, while creators are incentivized to alter their contents such that it is recommended more frequently. We define a model, called user-creator feature dynamics, to capture the dual influences of recommender systems. We prove that a recommender system with dual influence is guaranteed to polarize, causing diversity loss in the system. We then investigate, both theoretically and empirically, approaches for mitigating polarization and promoting diversity in recommender systems. Unexpectedly, we find that common diversity-promoting approaches do not work in the presence of dual influence, while relevancy-optimizing methods like top-$k$ recommendation can prevent polarization and improve diversity of the system.
Authors: Guan Li, Yang Liu, Guihua Shan, Shiyu Cheng, Weiqun Cao, Junpeng Wang, Ko-Chih Wang
Abstract: Numerical simulation serves as a cornerstone in scientific modeling, yet the process of fine-tuning simulation parameters poses significant challenges. Conventionally, parameter adjustment relies on extensive numerical simulations, data analysis, and expert insights, resulting in substantial computational costs and low efficiency. The emergence of deep learning in recent years has provided promising avenues for more efficient exploration of parameter spaces. However, existing approaches often lack intuitive methods for precise parameter adjustment and optimization. To tackle these challenges, we introduce ParamsDrag, a model that facilitates parameter space exploration through direct interaction with visualizations. Inspired by DragGAN, our ParamsDrag model operates in three steps. First, the generative component of ParamsDrag generates visualizations based on the input simulation parameters. Second, by directly dragging structure-related features in the visualizations, users can intuitively understand the controlling effect of different parameters. Third, with the understanding from the earlier step, users can steer ParamsDrag to produce dynamic visual outcomes. Through experiments conducted on real-world simulations and comparisons with state-of-the-art deep learning-based approaches, we demonstrate the efficacy of our solution.
Authors: Meng Zhang, Jie Sun, Qinghao Hu, Peng Sun, Zeke Wang, Yonggang Wen, Tianwei Zhang
Abstract: Graph Transformer is a new architecture that surpasses GNNs in graph learning. While there emerge inspiring algorithm advancements, their practical adoption is still limited, particularly on real-world graphs involving up to millions of nodes. We observe existing graph transformers fail on large-scale graphs mainly due to heavy computation, limited scalability and inferior model quality. Motivated by these observations, we propose TorchGT, the first efficient, scalable, and accurate graph transformer training system. TorchGT optimizes training at different levels. At algorithm level, by harnessing the graph sparsity, TorchGT introduces a Dual-interleaved Attention which is computation-efficient and accuracy-maintained. At runtime level, TorchGT scales training across workers with a communication-light Cluster-aware Graph Parallelism. At kernel level, an Elastic Computation Reformation further optimizes the computation by reducing memory access latency in a dynamic way. Extensive experiments demonstrate that TorchGT boosts training by up to 62.7x and supports graph sequence lengths of up to 1M.
Authors: Shuche Wang, Vincent Y. F. Tan
Abstract: Distributed gradient descent algorithms have come to the fore in modern machine learning, especially in parallelizing the handling of large datasets that are distributed across several workers. However, scant attention has been paid to analyzing the behavior of distributed gradient descent algorithms in the presence of adversarial corruptions instead of random noise. In this paper, we formulate a novel problem in which adversarial corruptions are present in a distributed learning system. We show how to use ideas from (lazy) mirror descent to design a corruption-tolerant distributed optimization algorithm. Extensive convergence analysis for (strongly) convex loss functions is provided for different choices of the stepsize. We carefully optimize the stepsize schedule to accelerate the convergence of the algorithm, while at the same time amortizing the effect of the corruption over time. Experiments based on linear regression, support vector classification, and softmax classification on the MNIST dataset corroborate our theoretical findings.
Authors: Shohreh Deldari, Mohammad Goudarzi, Aditya Joshi, Arash Shaghaghi, Simon Finn, Flora D. Salim, Sanjay Jha
Abstract: In the age of information overload, professionals across various fields face the challenge of navigating vast amounts of documentation and ever-evolving standards. Ensuring compliance with standards, regulations, and contractual obligations is a critical yet complex task across various professional fields. We propose a versatile conversational AI assistant framework designed to facilitate compliance checking on the go, in diverse domains, including but not limited to network infrastructure, legal contracts, educational standards, environmental regulations, and government policies. By leveraging retrieval-augmented generation using large language models, our framework automates the review, indexing, and retrieval of relevant, context-aware information, streamlining the process of verifying adherence to established guidelines and requirements. This AI assistant not only reduces the manual effort involved in compliance checks but also enhances accuracy and efficiency, supporting professionals in maintaining high standards of practice and ensuring regulatory compliance in their respective fields. We propose and demonstrate AuditNet, the first conversational AI security assistant designed to assist IoT network security experts by providing instant access to security standards, policies, and regulations.
Authors: Mulham Fawakherji, Vincenzo Suriani, Daniele Nardi, Domenico Daniele Bloisi
Abstract: The use of deep learning methods for precision farming is gaining increasing interest. However, collecting training data in this application field is particularly challenging and costly due to the need of acquiring information during the different growing stages of the cultivation of interest. In this paper, we present a method for data augmentation that uses two GANs to create artificial images to augment the training data. To obtain a higher image quality, instead of re-creating the entire scene, we take original images and replace only the patches containing objects of interest with artificial ones containing new objects with different shapes and styles. In doing this, we take into account both the foreground (i.e., crop samples) and the background (i.e., the soil) of the patches. Quantitative experiments, conducted on publicly available datasets, demonstrate the effectiveness of the proposed approach. The source code and data discussed in this work are available as open source.
Authors: Zhengyuan Xie, Haiquan Lu, Jia-wen Xiao, Enguang Wang, Le Zhang, Xialei Liu
Abstract: Class incremental semantic segmentation aims to preserve old knowledge while learning new tasks, however, it is impeded by catastrophic forgetting and background shift issues. Prior works indicate the pivotal importance of initializing new classifiers and mainly focus on transferring knowledge from the background classifier or preparing classifiers for future classes, neglecting the flexibility and variance of new classifiers. In this paper, we propose a new classifier pre-tuning~(NeST) method applied before the formal training process, learning a transformation from old classifiers to generate new classifiers for initialization rather than directly tuning the parameters of new classifiers. Our method can make new classifiers align with the backbone and adapt to the new data, preventing drastic changes in the feature extractor when learning new classes. Besides, we design a strategy considering the cross-task class similarity to initialize matrices used in the transformation, helping achieve the stability-plasticity trade-off. Experiments on Pascal VOC 2012 and ADE20K datasets show that the proposed strategy can significantly improve the performance of previous methods. The code is available at \url{https://github.com/zhengyuan-xie/ECCV24_NeST}.
Authors: Linlan Huang, Xusheng Cao, Haori Lu, Xialei Liu
Abstract: Class-incremental learning is a challenging problem, where the goal is to train a model that can classify data from an increasing number of classes over time. With the advancement of vision-language pre-trained models such as CLIP, they demonstrate good generalization ability that allows them to excel in class-incremental learning with completely frozen parameters. However, further adaptation to downstream tasks by simply fine-tuning the model leads to severe forgetting. Most existing works with pre-trained models assume that the forgetting of old classes is uniform when the model acquires new knowledge. In this paper, we propose a method named Adaptive Representation Adjustment and Parameter Fusion (RAPF). During training for new data, we measure the influence of new classes on old ones and adjust the representations, using textual features. After training, we employ a decomposed parameter fusion to further mitigate forgetting during adapter module fine-tuning. Experiments on several conventional benchmarks show that our method achieves state-of-the-art results. Our code is available at \url{https://github.com/linlany/RAPF}.
Authors: Henry Addison, Elizabeth Kendon, Suman Ravuri, Laurence Aitchison, Peter AG Watson
Abstract: High-resolution climate simulations are very valuable for understanding climate change impacts and planning adaptation measures. This has motivated use of regional climate models at sufficiently fine resolution to capture important small-scale atmospheric processes, such as convective storms. However, these regional models have very high computational costs, limiting their applicability. We present CPMGEM, a novel application of a generative machine learning model, a diffusion model, to skilfully emulate precipitation simulations from such a high-resolution model over England and Wales at much lower cost. This emulator enables stochastic generation of high-resolution (8.8km), daily-mean precipitation samples conditioned on coarse-resolution (60km) weather states from a global climate model. The output is fine enough for use in applications such as flood inundation modelling. The emulator produces precipitation predictions with realistic intensities and spatial structures and captures most of the 21st century climate change signal. We show evidence that the emulator has skill for extreme events up to and including 1-in-100 year intensities. Potential applications include producing high-resolution precipitation predictions for large-ensemble climate simulations and downscaling different climate models and climate change scenarios to better sample uncertainty in climate changes at local-scale.
Authors: Julian Gerstenberg, Ralph Neininger, Denis Spiegel
Abstract: We introduce a novel class of algorithms to efficiently approximate the unknown return distributions in policy evaluation problems from distributional reinforcement learning (DRL). The proposed distributional dynamic programming algorithms are suitable for underlying Markov decision processes (MDPs) having an arbitrary probabilistic reward mechanism, including continuous reward distributions with unbounded support being potentially heavy-tailed. For a plain instance of our proposed class of algorithms we prove error bounds, both within Wasserstein and Kolmogorov--Smirnov distances. Furthermore, for return distributions having probability density functions the algorithms yield approximations for these densities; error bounds are given within supremum norm. We introduce the concept of quantile-spline discretizations to come up with algorithms showing promising results in simulation experiments. While the performance of our algorithms can rigorously be analysed they can be seen as universal black box algorithms applicable to a large class of MDPs. We also derive new properties of probability metrics commonly used in DRL on which our quantitative analysis is based.
Authors: Giulia Vannucci, Roberta Siciliano, Andrea Saltelli
Abstract: The present work provides an application of Global Sensitivity Analysis to supervised machine learning methods such as Random Forests. These methods act as black boxes, selecting features in high--dimensional data sets as to provide accurate classifiers in terms of prediction when new data are fed into the system. In supervised machine learning, predictors are generally ranked by importance based on their contribution to the final prediction. Global Sensitivity Analysis is primarily used in mathematical modelling to investigate the effect of the uncertainties of the input variables on the output. We apply it here as a novel way to rank the input features by their importance to the explainability of the data generating process, shedding light on how the response is determined by the dependence structure of its predictors. A simulation study shows that our proposal can be used to explore what advances can be achieved either in terms of efficiency, explanatory ability, or simply by way of confirming existing results.
Authors: Zhiyuan Yang, Bo Zhang, Yufei Shi, Ningze Zhong, Johnathan Loh, Huihui Fang, Yanwu Xu, Si Yong Yeo
Abstract: Glaucoma is one of the leading causes of vision impairment. Digital imaging techniques, such as color fundus photography (CFP) and optical coherence tomography (OCT), provide quantitative and noninvasive methods for glaucoma diagnosis. Recently, in the field of computer-aided glaucoma diagnosis, multi-modality methods that integrate the CFP and OCT modalities have achieved greater diagnostic accuracy compared to single-modality methods. However, it remains challenging to extract reliable features due to the high similarity of medical images and the unbalanced multi-modal data distribution. Moreover, existing methods overlook the uncertainty estimation of different modalities, leading to unreliable predictions. To address these challenges, we propose a novel framework, namely ETSCL, which consists of a contrastive feature extraction stage and a decision-level fusion stage. Specifically, the supervised contrastive loss is employed to enhance the discriminative power in the feature extraction process, resulting in more effective features. In addition, we utilize the Frangi vesselness algorithm as a preprocessing step to incorporate vessel information to assist in the prediction. In the decision-level fusion stage, an evidence theory-based multi-modality classifier is employed to combine multi-source information with uncertainty estimation. Extensive experiments demonstrate that our method achieves state-of-the-art performance. The code is available at \url{https://github.com/master-Shix/ETSCL}.
Authors: Martin Menabue, Emanuele Frascaroli, Matteo Boschini, Lorenzo Bonicelli, Angelo Porrello, Simone Calderara
Abstract: The field of Continual Learning (CL) has inspired numerous researchers over the years, leading to increasingly advanced countermeasures to the issue of catastrophic forgetting. Most studies have focused on the single-class scenario, where each example comes with a single label. The recent literature has successfully tackled such a setting, with impressive results. Differently, we shift our attention to the multi-label scenario, as we feel it to be more representative of real-world open problems. In our work, we show that existing state-of-the-art CL methods fail to achieve satisfactory performance, thus questioning the real advance claimed in recent years. Therefore, we assess both old-style and novel strategies and propose, on top of them, an approach called Selective Class Attention Distillation (SCAD). It relies on a knowledge transfer technique that seeks to align the representations of the student network -- which trains continuously and is subject to forgetting -- with the teacher ones, which is pretrained and kept frozen. Importantly, our method is able to selectively transfer the relevant information from the teacher to the student, thereby preventing irrelevant information from harming the student's performance during online training. To demonstrate the merits of our approach, we conduct experiments on two different multi-label datasets, showing that our method outperforms the current state-of-the-art Continual Learning methods. Our findings highlight the importance of addressing the unique challenges posed by multi-label environments in the field of Continual Learning. The code of SCAD is available at https://github.com/aimagelab/SCAD-LOD-2024.
Authors: Nikolas Vitsakis, Amit Parekh, Ioannis Konstas
Abstract: Language models have been shown to reproduce underlying biases existing in their training data, which is the majority perspective by default. Proposed solutions aim to capture minority perspectives by either modelling annotator disagreements or grouping annotators based on shared metadata, both of which face significant challenges. We propose a framework that trains models without encoding annotator metadata, extracts latent embeddings informed by annotator behaviour, and creates clusters of similar opinions, that we refer to as voices. Resulting clusters are validated post-hoc via internal and external quantitative metrics, as well a qualitative analysis to identify the type of voice that each cluster represents. Our results demonstrate the strong generalisation capability of our framework, indicated by resulting clusters being adequately robust, while also capturing minority perspectives based on different demographic factors throughout two distinct datasets.
Authors: Alexandre d'Hooge (LaBRI, SCRIME), Louis Bigo (LaBRI, SCRIME), Ken D\'eguernel, Nicolas Martin
Abstract: Chord diagrams are used by guitar players to show where and how to play a chord on the fretboard. They are useful to beginners learning chords or for sharing the hand positions required to play a song.However, the diagrams presented on guitar learning toolsare usually selected from an existing databaseand rarely represent the actual positions used by performers.In this paper, we propose a tool which suggests a chord diagram for achord label,taking into account the diagram of the previous chord.Based on statistical analysis of the DadaGP and mySongBook datasets, we show that some chord diagrams are over-represented in western popular musicand that some chords can be played in more than 20 different ways.We argue that taking context into account can improve the variety and the quality of chord diagram suggestion, and compare this approach with a model taking only the current chord label into account.We show that adding previous context improves the F1-score on this task by up to 27% and reduces the propensity of the model to suggest standard open chords.We also define the notion of texture in the context of chord diagrams andshow through a variety of metrics that our model improves textureconsistencywith the previous diagram.
Authors: Xinzhou Jin, Jintang Li, Liang Chen, Chenyun Yu, Yuanzhen Xie, Tao Xie, Chengxiang Zhuo, Zang Li, Zibin Zheng
Abstract: Graph neural networks (GNNs) have recently emerged as an effective approach to model neighborhood signals in collaborative filtering. Towards this research line, graph contrastive learning (GCL) demonstrates robust capabilities to address the supervision label shortage issue through generating massive self-supervised signals. Despite its effectiveness, GCL for recommendation suffers seriously from two main challenges: i) GCL relies on graph augmentation to generate semantically different views for contrasting, which could potentially disrupt key information and introduce unwanted noise; ii) current works for GCL primarily focus on contrasting representations using sophisticated networks architecture (usually deep) to capture high-order interactions, which leads to increased computational complexity and suboptimal training efficiency. To this end, we propose L2CL, a principled Layer-to-Layer Contrastive Learning framework that contrasts representations from different layers. By aligning the semantic similarities between different layers, L2CL enables the learning of complex structural relationships and gets rid of the noise perturbation in stochastic data augmentation. Surprisingly, we find that L2CL, using only one-hop contrastive learning paradigm, is able to capture intrinsic semantic structures and improve the quality of node representation, leading to a simple yet effective architecture. We also provide theoretical guarantees for L2CL in minimizing task-irrelevant information. Extensive experiments on five real-world datasets demonstrate the superiority of our model over various state-of-the-art collaborative filtering methods. Our code is available at https://github.com/downeykking/L2CL.
Authors: Abdulrahman Ibraheem, Narsis Kiani, Jesper Tegner
Abstract: This paper studies the usefulness of incorporating path information in predicting chemical properties from molecular graphs, in the domain of QSAR (Quantitative Structure-Activity Relationship). Towards this, we developed a GNN-style model which can be toggled to operate in one of two modes: a non-degenerate mode which incorporates path information, and a degenerate mode which leaves out path information. Thus, by comparing the performance of the non-degenerate mode versus the degenerate mode on relevant QSAR datasets, we were able to directly assess the significance of path information on those datasets. Our results corroborate previous works, by suggesting that the usefulness of path information is datasetdependent. Unlike previous studies however, we took the very first steps towards building a model that could predict upfront whether or not path information would be useful for a given dataset at hand. Moreover, we also found that, albeit its simplicity, the degenerate mode of our model yielded rather surprising results, which outperformed more sophisticated SOTA models in certain cases.
Authors: Tiago Dias, Eva Maia, Isabel Pra\c{c}a
Abstract: Software's pervasive impact and increasing reliance in the era of digital transformation raise concerns about vulnerabilities, emphasizing the need for software security. Fuzzy testing is a dynamic analysis software testing technique that consists of feeding faulty input data to a System Under Test (SUT) and observing its behavior. Specifically regarding black-box RESTful API testing, recent literature has attempted to automate this technique using heuristics to perform the input search and using the HTTP response status codes for classification. However, most approaches do not keep track of code coverage, which is important to validate the solution. This work introduces a black-box RESTful API fuzzy testing tool that employs Reinforcement Learning (RL) for vulnerability detection. The fuzzer operates via the OpenAPI Specification (OAS) file and a scenarios file, which includes information to communicate with the SUT and the sequences of functionalities to test, respectively. To evaluate its effectiveness, the tool was tested on the Petstore API. The tool found a total of six unique vulnerabilities and achieved 55\% code coverage.
Authors: Vadim Borisov, Richard H. Schreiber
Abstract: The tremendous success of chat-based AI systems like ChatGPT, Claude, and Gemini stems from Large Language Models (LLMs) trained on vast amount of datasets. However, acquiring high-quality, diverse, and ethically sourced training data remains a significant challenge. We introduce the Open Artificial Knowledge (OAK) dataset, a large-scale resource of over 500 million tokens (at the moment of writing) designed to address this issue. OAK leverages an ensemble of state-of-the-art LLMs, including GPT4o, LLaMa3-70B, LLaMa3-8B, Mixtral-8x7B, Gemma-7B, and Gemma-2-9B , to generate high-quality text across diverse domains, guided by Wikipedia's main categories. Our methodology ensures broad knowledge coverage while maintaining coherence and factual accuracy. The OAK dataset aims to foster the development of more capable and aligned language models while addressing critical issues of data scarcity and privacy in LLM training, and it is freely available on www.oakdataset.org.
Authors: Jos\'e Gon\c{c}alves, Tiago Dias, Eva Maia, Isabel Pra\c{c}a
Abstract: In recent years, code security has become increasingly important, especially with the rise of interconnected technologies. Detecting vulnerabilities early in the software development process has demonstrated numerous benefits. Consequently, the scientific community started using machine learning for automated detection of source code vulnerabilities. This work explores and refines the CVEFixes dataset, which is commonly used to train models for code-related tasks, specifically the C/C++ subset. To this purpose, the Source Code Processing Engine (SCoPE), a framework composed of strategized techniques that can be used to reduce the size and normalize C/C++ functions is presented. The output generated by SCoPE was used to create a new version of CVEFixes. This refined dataset was then employed in a feature representation analysis to assess the effectiveness of the tool's code processing techniques, consisting of fine-tuning three pre-trained LLMs for software vulnerability detection. The results show that SCoPE successfully helped to identify 905 duplicates within the evaluated subset. The LLM results corroborate with the literature regarding their suitability for software vulnerability detection, with the best model achieving 53% F1-score.
Authors: Vaishnavi Kasuluru, Luis Blanco, Engin Zeydan
Abstract: Unlike other single-point Artificial Intelligence (AI)-based prediction techniques, such as Long-Short Term Memory (LSTM), probabilistic forecasting techniques (e.g., DeepAR and Transformer) provide a range of possible outcomes and associated probabilities that enable decision makers to make more informed and robust decisions. At the same time, the architecture of Open RAN has emerged as a revolutionary approach for mobile networks, aiming at openness, interoperability and innovation in the ecosystem of RAN. In this paper, we propose the use of probabilistic forecasting techniques as a radio App (rApp) within the Open RAN architecture. We investigate and compare different probabilistic and single-point forecasting methods and algorithms to estimate the utilization and resource demands of Physical Resource Blocks (PRBs) of cellular base stations. Through our evaluations, we demonstrate the numerical advantages of probabilistic forecasting techniques over traditional single-point forecasting methods and show that they are capable of providing more accurate and reliable estimates. In particular, DeepAR clearly outperforms single-point forecasting techniques such as LSTM and Seasonal-Naive (SN) baselines and other probabilistic forecasting techniques such as Simple-Feed-Forward (SFF) and Transformer neural networks.
Authors: Vaishnavi Kasuluru, Luis Blanco, Engin Zeydan, Albert Bel, Angelos Antonopoulos
Abstract: The need for intelligent and efficient resource provisioning for the productive management of resources in real-world scenarios is growing with the evolution of telecommunications towards the 6G era. Technologies such as Open Radio Access Network (O-RAN) can help to build interoperable solutions for the management of complex systems. Probabilistic forecasting, in contrast to deterministic single-point estimators, can offer a different approach to resource allocation by quantifying the uncertainty of the generated predictions. This paper examines the cloud-native aspects of O-RAN together with the radio App (rApp) deployment options. The integration of probabilistic forecasting techniques as a rApp in O-RAN is also emphasized, along with case studies of real-world applications. Through a comparative analysis of forecasting models using the error metric, we show the advantages of Deep Autoregressive Recurrent network (DeepAR) over other deterministic probabilistic estimators. Furthermore, the simplicity of Simple-Feed-Forward (SFF) leads to a fast runtime but does not capture the temporal dependencies of the input data. Finally, we present some aspects related to the practical applicability of cloud-native O-RAN with probabilistic forecasting.
Authors: Vaishnavi Kasuluru, Luis Blanco, Cristian J. Vaca-Rubio, Engin Zeydan
Abstract: The transition to sustainable Open Radio Access Network (O-RAN) architectures brings new challenges for resource management, especially in predicting the utilization of Physical Resource Block (PRB)s. In this paper, we propose a novel approach to characterize the PRB load using probabilistic forecasting techniques. First, we provide background information on the O-RAN architecture and components and emphasize the importance of energy/power consumption models for sustainable implementations. The problem statement highlights the need for accurate PRB load prediction to optimize resource allocation and power efficiency. We then investigate probabilistic forecasting techniques, including Simple-Feed-Forward (SFF), DeepAR, and Transformers, and discuss their likelihood model assumptions. The simulation results show that DeepAR estimators predict the PRBs with less uncertainty and effectively capture the temporal dependencies in the dataset compared to SFF- and Transformer-based models, leading to power savings. Different percentile selections can also increase power savings, but at the cost of over-/under provisioning. At the same time, the performance of the Long-Short Term Memory (LSTM) is shown to be inferior to the probabilistic estimators with respect to all error metrics. Finally, we outline the importance of probabilistic, prediction-based characterization for sustainable O-RAN implementations and highlight avenues for future research.
Authors: Tang Li, Mengmeng Ma, Xi Peng
Abstract: Large pre-trained Vision-Language Models (VLMs) have become ubiquitous foundational components of other models and downstream tasks. Although powerful, our empirical results reveal that such models might not be able to identify fine-grained concepts. Specifically, the explanations of VLMs with respect to fine-grained concepts are entangled and mislocalized. To address this issue, we propose to DisEntAngle and Localize (DEAL) the concept-level explanations for VLMs without human annotations. The key idea is encouraging the concept-level explanations to be distinct while maintaining consistency with category-level explanations. We conduct extensive experiments and ablation studies on a wide range of benchmark datasets and vision-language models. Our empirical results demonstrate that the proposed method significantly improves the concept-level explanations of the model in terms of disentanglability and localizability. Surprisingly, the improved explainability alleviates the model's reliance on spurious correlations, which further benefits the prediction accuracy.
Authors: Swarnadeep Saha, Archiki Prasad, Justin Chih-Yao Chen, Peter Hase, Elias Stengel-Eskin, Mohit Bansal
Abstract: Language models can be used to solve long-horizon planning problems in two distinct modes: a fast 'System-1' mode, directly generating plans without any explicit search or backtracking, and a slow 'System-2' mode, planning step-by-step by explicitly searching over possible actions. While System-2 is typically more effective, it is also more computationally expensive, making it infeasible for long plans or large action spaces. Moreover, isolated System-1 or 2 ignores the user's end goals, failing to provide ways to control the model's behavior. To this end, we propose the System-1.x Planner, a controllable planning framework with LLMs that is capable of generating hybrid plans and balancing between the two planning modes based on the difficulty of the problem at hand. System-1.x consists of (i) a controller, (ii) a System-1 Planner, and (iii) a System-2 Planner. Based on a user-specified hybridization factor (x) governing the mixture between System-1 and 2, the controller decomposes a problem into sub-goals, and classifies them as easy or hard to be solved by either System-1 or 2, respectively. We fine-tune all three components on top of a single base LLM, requiring only search traces as supervision. Experiments with two diverse planning tasks -- Maze Navigation and Blocksworld -- show that our System-1.x Planner outperforms a System-1 Planner, a System-2 Planner trained to approximate A* search, and also a symbolic planner (A*). We demonstrate the following key properties of our planner: (1) controllability: increasing the hybridization factor (e.g., System-1.75 vs 1.5) performs more search, improving performance, (2) flexibility: by building a neuro-symbolic variant with a neural System-1 and a symbolic System-2, we can use existing symbolic methods, and (3) generalizability: by being able to learn from different search algorithms, our method is robust to the choice of search algorithm.
Authors: HamidReza Imani, Abdolah Amirany, Tarek El-Ghazawi
Abstract: The increasing demand for deploying large Mixture-of-Experts (MoE) models in resource-constrained environments necessitates efficient approaches to address their high memory and computational requirements challenges. Moreover, given that tasks come in different user-defined constraints and the available resources change over time in multi-tenant environments, it is necessary to design an approach which provides a flexible configuration space. This paper presents an adaptive serving approach for the efficient deployment of MoE models, capitalizing on partial quantization of the experts. By dynamically determining the number of quantized experts and their distribution across CPU and GPU, our approach explores the Pareto frontier and offers a fine-grained range of configurations for tuning throughput and model quality. Our evaluation on an NVIDIA A100 GPU using a Mixtral 8x7B MoE model for three language modelling benchmarks demonstrates that the throughput of token generation can be adjusted from 0.63 to 13.00 token per second. This enhancement comes with a marginal perplexity increase of 2.62 to 2.80, 6.48 to 7.24, and 3.24 to 3.53 for WikiText2, PTB, and C4 datasets respectively under maximum quantization. These results highlight the practical applicability of our approach in dynamic and accuracy-sensitive applications where both memory usage and output quality are important.
Authors: Peng Xu, Wei Ping, Xianchao Wu, Zihan Liu, Mohammad Shoeybi, Bryan Catanzaro
Abstract: In this work, we introduce ChatQA 2, a Llama3-based model designed to bridge the gap between open-access LLMs and leading proprietary models (e.g., GPT-4-Turbo) in long-context understanding and retrieval-augmented generation (RAG) capabilities. These two capabilities are essential for LLMs to process large volumes of information that cannot fit into a single prompt and are complementary to each other, depending on the downstream tasks and computational budgets. We present a detailed continued training recipe to extend the context window of Llama3-70B-base from 8K to 128K tokens, along with a three-stage instruction tuning process to enhance the model's instruction-following, RAG performance, and long-context understanding capabilities. Our results demonstrate that the Llama3-ChatQA-2-70B model achieves accuracy comparable to GPT-4-Turbo-2024-0409 on many long-context understanding tasks and surpasses it on the RAG benchmark. Interestingly, we find that the state-of-the-art long-context retriever can alleviate the top-k context fragmentation issue in RAG, further improving RAG-based results for long-context understanding tasks. We also provide extensive comparisons between RAG and long-context solutions using state-of-the-art long-context LLMs.
Authors: Sukrut Rao, Sweta Mahajan, Moritz B\"ohle, Bernt Schiele
Abstract: Concept Bottleneck Models (CBMs) have recently been proposed to address the 'black-box' problem of deep neural networks, by first mapping images to a human-understandable concept space and then linearly combining concepts for classification. Such models typically require first coming up with a set of concepts relevant to the task and then aligning the representations of a feature extractor to map to these concepts. However, even with powerful foundational feature extractors like CLIP, there are no guarantees that the specified concepts are detectable. In this work, we leverage recent advances in mechanistic interpretability and propose a novel CBM approach -- called Discover-then-Name-CBM (DN-CBM) -- that inverts the typical paradigm: instead of pre-selecting concepts based on the downstream classification task, we use sparse autoencoders to first discover concepts learnt by the model, and then name them and train linear probes for classification. Our concept extraction strategy is efficient, since it is agnostic to the downstream task, and uses concepts already known to the model. We perform a comprehensive evaluation across multiple datasets and CLIP architectures and show that our method yields semantically meaningful concepts, assigns appropriate names to them that make them easy to interpret, and yields performant and interpretable CBMs. Code available at https://github.com/neuroexplicit-saar/discover-then-name.
URLs: https://github.com/neuroexplicit-saar/discover-then-name.
Authors: Elad Hazan, Karan Singh
Abstract: This text presents an introduction to an emerging paradigm in control of dynamical systems and differentiable reinforcement learning called online nonstochastic control. The new approach applies techniques from online convex optimization and convex relaxations to obtain new methods with provable guarantees for classical settings in optimal and robust control. The primary distinction between online nonstochastic control and other frameworks is the objective. In optimal control, robust control, and other control methodologies that assume stochastic noise, the goal is to perform comparably to an offline optimal strategy. In online nonstochastic control, both the cost functions as well as the perturbations from the assumed dynamical model are chosen by an adversary. Thus the optimal policy is not defined a priori. Rather, the target is to attain low regret against the best policy in hindsight from a benchmark class of policies. This objective suggests the use of the decision making framework of online convex optimization as an algorithmic methodology. The resulting methods are based on iterative mathematical optimization algorithms, and are accompanied by finite-time regret and computational complexity guarantees.
Authors: Luca Thiede, Chong Sun, Al\'an Aspuru-Guzik
Abstract: An efficient and expressive wavefunction ansatz is key to scalable solutions for complex many-body electronic structures. While Slater determinants are predominantly used for constructing antisymmetric electronic wavefunction ans\"{a}tze, this construction can result in limited expressiveness when the targeted wavefunction is highly complex. In this work, we introduce Waveflow, an innovative framework for learning many-body fermionic wavefunctions using boundary-conditioned normalizing flows. Instead of relying on Slater determinants, Waveflow imposes antisymmetry by defining the fundamental domain of the wavefunction and applying necessary boundary conditions. A key challenge in using normalizing flows for this purpose is addressing the topological mismatch between the prior and target distributions. We propose using O-spline priors and I-spline bijections to handle this mismatch, which allows for flexibility in the node number of the distribution while automatically maintaining its square-normalization property. We apply Waveflow to a one-dimensional many-electron system, where we variationally minimize the system's energy using variational quantum Monte Carlo (VQMC). Our experiments demonstrate that Waveflow can effectively resolve topological mismatches and faithfully learn the ground-state wavefunction.
Authors: Xiao Li, Sheng Liu, Jinxin Zhou, Xinyu Lu, Carlos Fernandez-Granda, Zhihui Zhu, Qing Qu
Abstract: With the ever-increasing complexity of large-scale pre-trained models coupled with a shortage of labeled data for downstream training, transfer learning has become the primary approach in many fields, including natural language processing, computer vision, and multi-modal learning. Despite recent progress, the fine-tuning process for large-scale pre-trained models in vision still mostly relies on trial and error. This work investigates the relationship between neural collapse (NC) and transfer learning for classification problems. NC is an intriguing while prevalent phenomenon that has been recently discovered in terms of the final-layer features and linear classifiers of trained neural networks. Specifically, during the terminal phase of training, NC implies that the variability of the features within each class diminishes to zero, while the means of features between classes are maximally and equally distanced. In this work, we examine the NC attributes of pre-trained models on both downstream and source data for transfer learning, and we find strong correlation between feature collapse and downstream performance. In particular, we discovered a systematic pattern that emerges when linear probing pre-trained models on downstream training data: the more feature collapse of pre-trained models on downstream training data, the higher the transfer accuracy. Additionally, we also studied the relationship between NC and transfer accuracy on the source data. Moreover, these findings allow us to develop a principled, parameter-efficient fine-tuning method that employs skip-connection to induce the last-layer feature collapse on downstream data. Our proposed fine-tuning methods deliver good performances while reducing fine-tuning parameters by at least 90% and mitigating overfitting in situations especially when the downstream data is scarce.
Authors: Fabian Altekr\"uger, Paul Hagemann, Gabriele Steidl
Abstract: Conditional generative models became a very powerful tool to sample from Bayesian inverse problem posteriors. It is well-known in classical Bayesian literature that posterior measures are quite robust with respect to perturbations of both the prior measure and the negative log-likelihood, which includes perturbations of the observations. However, to the best of our knowledge, the robustness of conditional generative models with respect to perturbations of the observations has not been investigated yet. In this paper, we prove for the first time that appropriately learned conditional generative models provide robust results for single observations.
Authors: Jianan Chen, Vishwesh Ramanathan, Tony Xu, Anne L. Martel
Abstract: Machine learning models experience deteriorated performance when trained in the presence of noisy labels. This is particularly problematic for medical tasks, such as survival prediction, which typically face high label noise complexity with few clear-cut solutions. Inspired by the large fluctuations across folds in the cross-validation performance of survival analyses, we design Monte-Carlo experiments to show that such fluctuation could be caused by label noise. We propose two novel and straightforward label noise detection algorithms that effectively identify noisy examples by pinpointing the samples that more frequently contribute to inferior cross-validation results. We first introduce Repeated Cross-Validation (ReCoV), a parameter-free label noise detection algorithm that is robust to model choice. We further develop fastReCoV, a less robust but more tractable and efficient variant of ReCoV suitable for deep learning applications. Through extensive experiments, we show that ReCoV and fastReCoV achieve state-of-the-art label noise detection performance in a wide range of modalities, models and tasks, including survival analysis, which has yet to be addressed in the literature. Our code and data are publicly available at https://github.com/GJiananChen/ReCoV.
Authors: Mohammad Pezeshki, Diane Bouchacourt, Mark Ibrahim, Nicolas Ballas, Pascal Vincent, David Lopez-Paz
Abstract: Environment annotations are essential for the success of many out-of-distribution (OOD) generalization methods. Unfortunately, these are costly to obtain and often limited by human annotators' biases. To achieve robust generalization, it is essential to develop algorithms for automatic environment discovery within datasets. Current proposals, which divide examples based on their training error, suffer from one fundamental problem. These methods introduce hyper-parameters and early-stopping criteria, which require a validation set with human-annotated environments, the very information subject to discovery. In this paper, we propose Cross-Risk-Minimization (XRM) to address this issue. XRM trains twin networks, each learning from one random half of the training data, while imitating confident held-out mistakes made by its sibling. XRM provides a recipe for hyper-parameter tuning, does not require early-stopping, and can discover environments for all training and validation data. Algorithms built on top of XRM environments achieve oracle worst-group-accuracy, addressing a long-standing challenge in OOD generalization. Code available at \url{https://github.com/facebookresearch/XRM}.
Authors: Jiaxu Liu, Xinping Yi, Xiaowei Huang
Abstract: Hyperbolic graph convolutional networks (HGCNs) have demonstrated significant potential in extracting information from hierarchical graphs. However, existing HGCNs are limited to shallow architectures due to the computational expense of hyperbolic operations and the issue of over-smoothing as depth increases. Although treatments have been applied to alleviate over-smoothing in GCNs, developing a hyperbolic solution presents distinct challenges since operations must be carefully designed to fit the hyperbolic nature. Addressing these challenges, we propose DeepHGCN, the first deep multi-layer HGCN architecture with dramatically improved computational efficiency and substantially reduced over-smoothing. DeepHGCN features two key innovations: (1) a novel hyperbolic feature transformation layer that enables fast and accurate linear mappings, and (2) techniques such as hyperbolic residual connections and regularization for both weights and features, facilitated by an efficient hyperbolic midpoint method. Extensive experiments demonstrate that DeepHGCN achieves significant improvements in link prediction and node classification tasks compared to both Euclidean and shallow hyperbolic GCN variants.
Authors: Arshia Soltani Moakhar, Eugenia Iofinova, Elias Frantar, Dan Alistarh
Abstract: It is known that sparsity can improve interpretability for deep neural networks. However, existing methods in the area either require networks that are pre-trained with sparsity constraints, or impose sparsity after the fact, altering the network's general behavior. In this paper, we demonstrate, for the first time, that sparsity can instead be incorporated into the interpretation process itself, as a sample-specific preprocessing step. Unlike previous work, this approach, which we call SPADE, does not place constraints on the trained model and does not affect its behavior during inference on the sample. Given a trained model and a target sample, SPADE uses sample-targeted pruning to provide a "trace" of the network's execution on the sample, reducing the network to the most important connections prior to computing an interpretation. We demonstrate that preprocessing with SPADE significantly increases the accuracy of image saliency maps across several interpretability methods. Additionally, SPADE improves the usefulness of neuron visualizations, aiding humans in reasoning about network behavior. Our code is available at https://github.com/IST-DASLab/SPADE.
Authors: Patrick Chao, Alexander Robey, Edgar Dobriban, Hamed Hassani, George J. Pappas, Eric Wong
Abstract: There is growing interest in ensuring that large language models (LLMs) align with human values. However, the alignment of such models is vulnerable to adversarial jailbreaks, which coax LLMs into overriding their safety guardrails. The identification of these vulnerabilities is therefore instrumental in understanding inherent weaknesses and preventing future misuse. To this end, we propose Prompt Automatic Iterative Refinement (PAIR), an algorithm that generates semantic jailbreaks with only black-box access to an LLM. PAIR -- which is inspired by social engineering attacks -- uses an attacker LLM to automatically generate jailbreaks for a separate targeted LLM without human intervention. In this way, the attacker LLM iteratively queries the target LLM to update and refine a candidate jailbreak. Empirically, PAIR often requires fewer than twenty queries to produce a jailbreak, which is orders of magnitude more efficient than existing algorithms. PAIR also achieves competitive jailbreaking success rates and transferability on open and closed-source LLMs, including GPT-3.5/4, Vicuna, and Gemini.
Authors: Luka Ribar, Ivan Chelombiev, Luke Hudlass-Galley, Charlie Blake, Carlo Luschi, Douglas Orr
Abstract: The computational difficulties of large language model (LLM) inference remain a significant obstacle to their widespread deployment. The need for many applications to support long input sequences and process them in large batches typically causes token-generation to be bottlenecked by data transfer. For this reason, we introduce SparQ Attention, a technique for increasing the inference throughput of LLMs by utilising memory bandwidth more efficiently within the attention layers, through selective fetching of the cached history. Our proposed technique can be applied directly to off-the-shelf LLMs during inference, without requiring any modification to the pre-training setup or additional fine-tuning. We show that SparQ Attention brings up to 8x savings in attention data transfers without substantial drops in accuracy, by evaluating Llama 2 and 3, Mistral, Gemma and Pythia models on a wide range of downstream tasks.
Authors: Yin Li, Yu Xiong, Wenxin Fan, Kai Wang, Qingqing Yu, Liping Si, Patrick van der Smagt, Jun Tang, Nutan Chen
Abstract: Objective: Subcutaneous Immunotherapy (SCIT) is the long-lasting causal treatment of allergic rhinitis (AR). How to enhance the adherence of patients to maximize the benefit of allergen immunotherapy (AIT) plays a crucial role in the management of AIT. This study aims to leverage novel machine learning models to precisely predict the risk of non-adherence of AR patients and related local symptom scores in three years SCIT. Methods: The research develops and analyzes two models, sequential latent-variable model (SLVM) of Stochastic Latent Actor-Critic (SLAC) and Long Short-Term Memory (LSTM) evaluating them based on scoring and adherence prediction capabilities. Results: Excluding the biased samples at the first time step, the predictive adherence accuracy of the SLAC models is from 60\% to 72\%, and for LSTM models, it is 66\% to 84\%, varying according to the time steps. The range of Root Mean Square Error (RMSE) for SLAC models is between 0.93 and 2.22, while for LSTM models it is between 1.09 and 1.77. Notably, these RMSEs are significantly lower than the random prediction error of 4.55. Conclusion: We creatively apply sequential models in the long-term management of SCIT with promising accuracy in the prediction of SCIT nonadherence in AR patients. While LSTM outperforms SLAC in adherence prediction, SLAC excels in score prediction for patients undergoing SCIT for AR. The state-action-based SLAC adds flexibility, presenting a novel and effective approach for managing long-term AIT.
Authors: Yifei He, Shiji Zhou, Guojun Zhang, Hyokun Yun, Yi Xu, Belinda Zeng, Trishul Chilimbi, Han Zhao
Abstract: Multi-task learning (MTL) considers learning a joint model for multiple tasks by optimizing a convex combination of all task losses. To solve the optimization problem, existing methods use an adaptive weight updating scheme, where task weights are dynamically adjusted based on their respective losses to prioritize difficult tasks. However, these algorithms face a great challenge whenever label noise is present, in which case excessive weights tend to be assigned to noisy tasks that have relatively large Bayes optimal errors, thereby overshadowing other tasks and causing performance to drop across the board. To overcome this limitation, we propose Multi-Task Learning with Excess Risks (ExcessMTL), an excess risk-based task balancing method that updates the task weights by their distances to convergence instead. Intuitively, ExcessMTL assigns higher weights to worse-trained tasks that are further from convergence. To estimate the excess risks, we develop an efficient and accurate method with Taylor approximation. Theoretically, we show that our proposed algorithm achieves convergence guarantees and Pareto stationarity. Empirically, we evaluate our algorithm on various MTL benchmarks and demonstrate its superior performance over existing methods in the presence of label noise. Our code is available at https://github.com/yifei-he/ExcessMTL.
Authors: Can Jin, Tong Che, Hongwu Peng, Yiyuan Li, Dimitris N. Metaxas, Marco Pavone
Abstract: Generalization remains a central challenge in machine learning. In this work, we propose Learning from Teaching (LoT), a novel regularization technique for deep neural networks to enhance generalization. Inspired by the human ability to capture concise and abstract patterns, we hypothesize that generalizable correlations are expected to be easier to imitate. LoT operationalizes this concept to improve generalization of the main model with auxiliary student learners. The student learners are trained by the main model and, in turn, provide feedback to help the main model capture more generalizable and imitable correlations. Our experimental results across several domains, including Computer Vision, Natural Language Processing, and methodologies like Reinforcement Learning, demonstrate that the introduction of LoT brings significant benefits compared to training models on the original dataset. The results suggest the effectiveness and efficiency of LoT in identifying generalizable information at the right scales while discarding spurious data correlations, thus making LoT a valuable addition to current machine learning. Code is available at https://github.com/jincan333/LoT.
Authors: Etienne Russeil, Fabr\'icio Olivetti de Fran\c{c}a, Konstantin Malanchev, Bogdan Burlacu, Emille E. O. Ishida, Marion Leroux, Cl\'ement Michelin, Guillaume Moinard, Emmanuel Gangler
Abstract: Symbolic regression (SR) searches for analytical expressions representing the relationship between a set of explanatory and response variables. Current SR methods assume a single dataset extracted from a single experiment. Nevertheless, frequently, the researcher is confronted with multiple sets of results obtained from experiments conducted with different setups. Traditional SR methods may fail to find the underlying expression since the parameters of each experiment can be different. In this work we present Multi-View Symbolic Regression (MvSR), which takes into account multiple datasets simultaneously, mimicking experimental environments, and outputs a general parametric solution. This approach fits the evaluated expression to each independent dataset and returns a parametric family of functions f(x; theta) simultaneously capable of accurately fitting all datasets. We demonstrate the effectiveness of MvSR using data generated from known expressions, as well as real-world data from astronomy, chemistry and economy, for which an a priori analytical expression is not available. Results show that MvSR obtains the correct expression more frequently and is robust to hyperparameters change. In real-world data, it is able to grasp the group behavior, recovering known expressions from the literature as well as promising alternatives, thus enabling the use of SR to a large range of experimental scenarios.
Authors: Yvonne Zhou, Mingyu Liang, Ivan Brugere, Dana Dachman-Soled, Danial Dervovic, Antigoni Polychroniadou, Min Wu
Abstract: The growing use of machine learning (ML) has raised concerns that an ML model may reveal private information about an individual who has contributed to the training dataset. To prevent leakage of sensitive data, we consider using differentially-private (DP), synthetic training data instead of real training data to train an ML model. A key desirable property of synthetic data is its ability to preserve the low-order marginals of the original distribution. Our main contribution comprises novel upper and lower bounds on the excess empirical risk of linear models trained on such synthetic data, for continuous and Lipschitz loss functions. We perform extensive experimentation alongside our theoretical results.
Authors: Jason Yoo, Yunpeng Liu, Frank Wood, Geoff Pleiss
Abstract: In online continual learning, a neural network incrementally learns from a non-i.i.d. data stream. Nearly all online continual learning methods employ experience replay to simultaneously prevent catastrophic forgetting and underfitting on past data. Our work demonstrates a limitation of this approach: neural networks trained with experience replay tend to have unstable optimization trajectories, impeding their overall accuracy. Surprisingly, these instabilities persist even when the replay buffer stores all previous training examples, suggesting that this issue is orthogonal to catastrophic forgetting. We minimize these instabilities through a simple modification of the optimization geometry. Our solution, Layerwise Proximal Replay (LPR), balances learning from new and replay data while only allowing for gradual changes in the hidden activation of past data. We demonstrate that LPR consistently improves replay-based online continual learning methods across multiple problem settings, regardless of the amount of available replay memory.
Authors: Jonas Beck, Nathanael Bosch, Michael Deistler, Kyra L. Kadhim, Jakob H. Macke, Philipp Hennig, Philipp Berens
Abstract: Ordinary differential equations (ODEs) are widely used to describe dynamical systems in science, but identifying parameters that explain experimental measurements is challenging. In particular, although ODEs are differentiable and would allow for gradient-based parameter optimization, the nonlinear dynamics of ODEs often lead to many local minima and extreme sensitivity to initial conditions. We therefore propose diffusion tempering, a novel regularization technique for probabilistic numerical methods which improves convergence of gradient-based parameter optimization in ODEs. By iteratively reducing a noise parameter of the probabilistic integrator, the proposed method converges more reliably to the true parameters. We demonstrate that our method is effective for dynamical systems of different complexity and show that it obtains reliable parameter estimates for a Hodgkin-Huxley model with a practically relevant number of parameters.
Authors: Adrian M\"uller, Pragnya Alatur, Volkan Cevher, Giorgia Ramponi, Niao He
Abstract: Constrained Markov decision processes (CMDPs) are a common way to model safety constraints in reinforcement learning. State-of-the-art methods for efficiently solving CMDPs are based on primal-dual algorithms. For these algorithms, all currently known regret bounds allow for error cancellations -- one can compensate for a constraint violation in one round with a strict constraint satisfaction in another. This makes the online learning process unsafe since it only guarantees safety for the final (mixture) policy but not during learning. As Efroni et al. (2020) pointed out, it is an open question whether primal-dual algorithms can provably achieve sublinear regret if we do not allow error cancellations. In this paper, we give the first affirmative answer. We first generalize a result on last-iterate convergence of regularized primal-dual schemes to CMDPs with multiple constraints. Building upon this insight, we propose a model-based primal-dual algorithm to learn in an unknown CMDP. We prove that our algorithm achieves sublinear regret without error cancellations.
Authors: Linan Yue, Qi Liu, Yichao Du, Li Wang, Weibo Gao, Yanqing An
Abstract: The remarkable success in neural networks provokes the selective rationalization. It explains the prediction results by identifying a small subset of the inputs sufficient to support them. Since existing methods still suffer from adopting the shortcuts in data to compose rationales and limited large-scale annotated rationales by human, in this paper, we propose a Shortcuts-fused Selective Rationalization (SSR) method, which boosts the rationalization by discovering and exploiting potential shortcuts. Specifically, SSR first designs a shortcuts discovery approach to detect several potential shortcuts. Then, by introducing the identified shortcuts, we propose two strategies to mitigate the problem of utilizing shortcuts to compose rationales. Finally, we develop two data augmentations methods to close the gap in the number of annotated rationales. Extensive experimental results on real-world datasets clearly validate the effectiveness of our proposed method.
Authors: R\'emi Nahon, Ivan Luiz De Moura Matos, Van-Tam Nguyen, Enzo Tartaglione
Abstract: Nowadays an ever-growing concerning phenomenon, the emergence of algorithmic biases that can lead to unfair models, emerges. Several debiasing approaches have been proposed in the realm of deep learning, employing more or less sophisticated approaches to discourage these models from massively employing these biases. However, a question emerges: is this extra complexity really necessary? Is a vanilla-trained model already embodying some ``unbiased sub-networks'' that can be used in isolation and propose a solution without relying on the algorithmic biases? In this work, we show that such a sub-network typically exists, and can be extracted from a vanilla-trained model without requiring additional training. We further validate that such specific architecture is incapable of learning a specific bias, suggesting that there are possible architectural countermeasures to the problem of biases in deep neural networks.
Authors: Wei Huang, Xingyu Zheng, Xudong Ma, Haotong Qin, Chengtao Lv, Hong Chen, Jie Luo, Xiaojuan Qi, Xianglong Liu, Michele Magno
Abstract: The LLaMA family has become one of the most powerful open-source Large Language Models (LLMs) and the popular LLM backbones of Multimodal Large Language Models (MLLMs), widely applied in Computer Vision (CV) and Natural Language Understanding (NLU) tasks. Notably, LLaMA3 models have recently been released and achieve impressive performance across various with super-large scale pre-training on over 15T tokens of data. Given the wide application of low-bit quantization for LLMs in resource-limited scenarios, we explore LLaMA3's capabilities when quantized to low bit-width. This exploration can potentially unveil new insights and challenges for low-bit quantization of LLaMA3 and other forthcoming LLMs, especially in addressing performance degradation problems that suffer in LLM compression. Specifically, we comprehensively evaluate the 10 existing post-training quantization and LoRA-finetuning methods of LLaMA3 on 1-8 bits and diverse datasets to reveal LLaMA3's low-bit quantization performance. To uncover the capabilities of low-bit quantized MLLM, we assessed the performance of the LLaMA3-based LLaVA-Next-8B model under 2-4 ultra-low bits with post-training quantization methods. Our experimental results indicate that LLaMA3 still suffers non-negligent degradation in linguistic and visual contexts, particularly under ultra-low bit widths. This highlights the significant performance gap under low bit-width that needs to be bridged in future developments. We expect that this empirical study will prove valuable in advancing future models, driving LLMs and MLLMs to achieve higher accuracy at lower bit to enhance practicality.
Authors: Zhengsen Xu, Jonathan Li, Linlin Xu
Abstract: Wildfires have significant impacts on global vegetation, wildlife, and humans. They destroy plant communities and wildlife habitats and contribute to increased emissions of carbon dioxide, nitrogen oxides, methane, and other pollutants. The prediction of wildfires relies on various independent variables combined with regression or machine learning methods. In this technical review, we describe the options for independent variables, data processing techniques, models, independent variables collinearity and importance estimation methods, and model performance evaluation metrics. First, we divide the independent variables into 4 aspects, including climate and meteorology conditions, socio-economical factors, terrain and hydrological features, and wildfire historical records. Second, preprocessing methods are described for different magnitudes, different spatial-temporal resolutions, and different formats of data. Third, the collinearity and importance evaluation methods of independent variables are also considered. Fourth, we discuss the application of statistical models, traditional machine learning models, and deep learning models in wildfire risk prediction. In this subsection, compared with other reviews, this manuscript particularly discusses the evaluation metrics and recent advancements in deep learning methods. Lastly, addressing the limitations of current research, this paper emphasizes the need for more effective deep learning time series forecasting algorithms, the utilization of three-dimensional data including ground and trunk fuel, extraction of more accurate historical fire point data, and improved model evaluation metrics.
Authors: Lorenzo Sani, Alex Iacob, Zeyu Cao, Bill Marino, Yan Gao, Tomas Paulik, Wanru Zhao, William F. Shen, Preslav Aleksandrov, Xinchi Qiu, Nicholas D. Lane
Abstract: Generative pre-trained large language models (LLMs) have demonstrated impressive performance over a wide range of tasks, thanks to the unprecedented amount of data they have been trained on. As established scaling laws indicate, LLMs' future performance improvement depends on the amount of computing and data sources they can leverage for pre-training. Federated learning (FL) has the potential to unleash the majority of the planet's data and computational resources, which are underutilized by the data-center-focused training methodology of current LLM practice. Our work presents a robust, flexible, reproducible FL approach that enables large-scale collaboration across institutions to train LLMs. We propose a scalable deployment system called Photon to enable the investigation and development of this new training paradigm for LLM pre-training. We show that Photon can be used by organizations interested in collaborating with their private data sources and computational resources for pre-training LLMs with billions of parameters. This paradigm would mobilize more computational and data resources while matching or potentially exceeding centralized performance. We further show the effectiveness of the federated training scales with model size and present our approach for training a billion-scale federated LLM using limited resources. Finally, we show that LLM training is highly resilient to the classical challenges of federated statistical and hardware heterogeneity. Furthermore, we show that convergence is robust to partial participation, opening the avenue for compute-efficient collaborative training. Photon will help data-rich actors to become the protagonists of LLMs pre-training instead of leaving the stage to compute-rich actors alone.
Authors: Yizhuo Chen, Chun-Fu Chen, Hsiang Hsu, Shaohan Hu, Marco Pistoia, Tarek Abdelzaher
Abstract: The growing richness of large-scale datasets has been crucial in driving the rapid advancement and wide adoption of machine learning technologies. The massive collection and usage of data, however, pose an increasing risk for people's private and sensitive information due to either inadvertent mishandling or malicious exploitation. Besides legislative solutions, many technical approaches have been proposed towards data privacy protection. However, they bear various limitations such as leading to degraded data availability and utility, or relying on heuristics and lacking solid theoretical bases. To overcome these limitations, we propose a formal information-theoretic definition for this utility-preserving privacy protection problem, and design a data-driven learnable data transformation framework that is capable of selectively suppressing sensitive attributes from target datasets while preserving the other useful attributes, regardless of whether or not they are known in advance or explicitly annotated for preservation. We provide rigorous theoretical analyses on the operational bounds for our framework, and carry out comprehensive experimental evaluations using datasets of a variety of modalities, including facial images, voice audio clips, and human activity motion sensor signals. Results demonstrate the effectiveness and generalizability of our method under various configurations on a multitude of tasks. Our code is available at https://github.com/jpmorganchase/MaSS.
Authors: Claudio Battiloro, Ege Karaismailo\u{g}lu, Mauricio Tec, George Dasoulas, Michelle Audirac, Francesca Dominici
Abstract: Graph neural networks excel at modeling pairwise interactions, but they cannot flexibly accommodate higher-order interactions and features. Topological deep learning (TDL) has emerged recently as a promising tool for addressing this issue. TDL enables the principled modeling of arbitrary multi-way, hierarchical higher-order interactions by operating on combinatorial topological spaces, such as simplicial or cell complexes, instead of graphs. However, little is known about how to leverage geometric features such as positions and velocities for TDL. This paper introduces E(n)-Equivariant Topological Neural Networks (ETNNs), which are E(n)-equivariant message-passing networks operating on combinatorial complexes, formal objects unifying graphs, hypergraphs, simplicial, path, and cell complexes. ETNNs incorporate geometric node features while respecting rotation and translation equivariance. Moreover, ETNNs are natively ready for settings with heterogeneous interactions. We provide a theoretical analysis to show the improved expressiveness of ETNNs over architectures for geometric graphs. We also show how several E(n) equivariant variants of TDL models can be directly derived from our framework. The broad applicability of ETNNs is demonstrated through two tasks of vastly different nature: i) molecular property prediction on the QM9 benchmark and ii) land-use regression for hyper-local estimation of air pollution with multi-resolution irregular geospatial data. The experiment results indicate that ETNNs are an effective tool for learning from diverse types of richly structured data, highlighting the benefits of principled geometric inductive bias.
Authors: Runquan Zhang, Jiawen Jiang, Xiaoping Shi
Abstract: Traditional survival analysis methods often struggle with complex time-dependent data,failing to capture and interpret dynamic characteristics adequately.This study aims to evaluate the performance of three long-sequence models,LSTM,Transformer,and Mamba,in analyzing recurrence event data and integrating them with the Cox proportional hazards model.This study integrates the advantages of deep learning models for handling long-sequence data with the Cox proportional hazards model to enhance the performance in analyzing recurrent events with dynamic time information.Additionally,this study compares the ability of different models to extract and utilize features from time-dependent clinical recurrence data.The LSTM-Cox model outperformed both the Transformer-Cox and Mamba-Cox models in prediction accuracy and model fit,achieving a Concordance index of up to 0.90 on the test set.Significant predictors of bladder cancer recurrence,such as treatment stop time,maximum tumor size at recurrence and recurrence frequency,were identified.The LSTM-Cox model aligned well with clinical outcomes,effectively distinguishing between high-risk and low-risk patient groups.This study demonstrates that the LSTM-Cox model is a robust and efficient method for recurrent data analysis and feature extraction,surpassing newer models like Transformer and Mamba.It offers a practical approach for integrating deep learning technologies into clinical risk prediction systems,thereby improving patient management and treatment outcomes.
Authors: Nathan Ng, Roger Grosse, Marzyeh Ghassemi
Abstract: Estimating the uncertainty of a model's prediction on a test point is a crucial part of ensuring reliability and calibration under distribution shifts. A minimum description length approach to this problem uses the predictive normalized maximum likelihood (pNML) distribution, which considers every possible label for a data point, and decreases confidence in a prediction if other labels are also consistent with the model and training data. In this work we propose IF-COMP, a scalable and efficient approximation of the pNML distribution that linearizes the model with a temperature-scaled Boltzmann influence function. IF-COMP can be used to produce well-calibrated predictions on test points as well as measure complexity in both labelled and unlabelled settings. We experimentally validate IF-COMP on uncertainty calibration, mislabel detection, and OOD detection tasks, where it consistently matches or beats strong baseline methods.
Authors: Tristan Cinquin, Robert Bamler
Abstract: Bayesian neural networks (BNN) promise to combine the predictive performance of neural networks with principled uncertainty modeling important for safety-critical systems and decision making. However, posterior uncertainty estimates depend on the choice of prior, and finding informative priors in weight-space has proven difficult. This has motivated variational inference (VI) methods that pose priors directly on the function generated by the BNN rather than on weights. In this paper, we address a fundamental issue with such function-space VI approaches pointed out by Burt et al. (2020), who showed that the objective function (ELBO) is negative infinite for most priors of interest. Our solution builds on generalized VI (Knoblauch et al., 2019) with the regularized KL divergence (Quang, 2019) and is, to the best of our knowledge, the first well-defined variational objective for function-space inference in BNNs with Gaussian process (GP) priors. Experiments show that our method incorporates the properties specified by the GP prior on synthetic and small real-world data sets, and provides competitive uncertainty estimates for regression, classification and out-of-distribution detection compared to BNN baselines with both function and weight-space priors.
Authors: Zhaoru Ke, Hang Yu, Jianguo Li, Haipeng Zhang
Abstract: Current directed graph embedding methods build upon undirected techniques but often inadequately capture directed edge information, leading to challenges such as: (1) Suboptimal representations for nodes with low in/out-degrees, due to the insufficient neighbor interactions; (2) Limited inductive ability for representing new nodes post-training; (3) Narrow generalizability, as training is overly coupled with specific tasks. In response, we propose DUPLEX, an inductive framework for complex embeddings of directed graphs. It (1) leverages Hermitian adjacency matrix decomposition for comprehensive neighbor integration, (2) employs a dual GAT encoder for directional neighbor modeling, and (3) features two parameter-free decoders to decouple training from particular tasks. DUPLEX outperforms state-of-the-art models, especially for nodes with sparse connectivity, and demonstrates robust inductive capability and adaptability across various tasks. The code is available at https://github.com/alipay/DUPLEX.
Authors: Lin Wang, Zhichao Wang, Xiaoying Tang
Abstract: The advent of large language models (LLMs) has revolutionized the deep learning paradigm, yielding impressive results across a wide array of tasks. However, the pre-training or fine-tuning of LLMs within a federated learning (FL) framework poses substantial challenges, including considerable computational and memory resource demands, as well as communication bottlenecks between servers and clients. Existing solutions either make the unrealistic assumption that the entire model is exchanged for training, or apply parameter-effective fine-tuning methods from centralized learning to train LLMs in FL which tend to underperform during training or fine-tuning stages due to the limited search subspace of parameter updating. In this paper, we introduce a novel method for the efficient training and fine-tuning of LLMs in FL, with minimal resource consumption. Our approach, termed FedCyBGD, utilizes Cycle Block Gradient Descent to periodically update the model. In particular, we design a compression scheme for FedCyBGD, aiming to further decrease the model download cost. It enables full parameter training in FL with only selected block updates and uploads, thereby reducing communication, computation, and memory costs. Our method achieves state-of-the-art performance for FL LLM training, while significantly reducing associated costs. Codes are provided here.
Authors: Anshuman Suri, Xiao Zhang, David Evans
Abstract: Membership inference attacks aim to infer whether an individual record was used to train a model, serving as a key tool for disclosure auditing. While such evaluations are useful to demonstrate risk, they are computationally expensive and often make strong assumptions about potential adversaries' access to models and training environments, and thus do not provide very tight bounds on leakage from potential attacks. We show how prior claims around black-box access being sufficient for optimal membership inference do not hold for most useful settings such as stochastic gradient descent, and that optimal membership inference indeed requires white-box access. We validate our findings with a new white-box inference attack IHA (Inverse Hessian Attack) that explicitly uses model parameters by taking advantage of computing inverse-Hessian vector products. Our results show that both audits and adversaries may be able to benefit from access to model parameters, and we advocate for further research into white-box methods for membership privacy auditing.
Authors: Tianyi Zhao, Jian Kang, Lu Cheng
Abstract: Graph Neural Networks (GNNs) excel in diverse tasks, yet their applications in high-stakes domains are often hampered by unreliable predictions. Although numerous uncertainty quantification methods have been proposed to address this limitation, they often lack \textit{rigorous} uncertainty estimates. This work makes the first attempt to introduce a distribution-free and model-agnostic uncertainty quantification approach to construct a predictive interval with a statistical guarantee for GNN-based link prediction. We term it as \textit{conformalized link prediction.} Our approach builds upon conformal prediction (CP), a framework that promises to construct statistically robust prediction sets or intervals. We first theoretically and empirically establish a permutation invariance condition for the application of CP in link prediction tasks, along with an exact test-time coverage. Leveraging the important structural information in graphs, we then identify a novel and crucial connection between a graph's adherence to the power law distribution and the efficiency of CP. This insight leads to the development of a simple yet effective sampling-based method to align the graph structure with a power law distribution prior to the standard CP procedure. Extensive experiments demonstrate that for conformalized link prediction, our approach achieves the desired marginal coverage while significantly improving the efficiency of CP compared to baseline methods.
Authors: Benjamin C. Koenig, Suyong Kim, Sili Deng
Abstract: Kolmogorov-Arnold networks (KANs) as an alternative to multi-layer perceptrons (MLPs) are a recent development demonstrating strong potential for data-driven modeling. This work applies KANs as the backbone of a neural ordinary differential equation (ODE) framework, generalizing their use to the time-dependent and temporal grid-sensitive cases often seen in dynamical systems and scientific machine learning applications. The proposed KAN-ODEs retain the flexible dynamical system modeling framework of Neural ODEs while leveraging the many benefits of KANs compared to MLPs, including higher accuracy and faster neural scaling, stronger interpretability and generalizability, and lower parameter counts. First, we quantitatively demonstrated these improvements in a comprehensive study of the classical Lotka-Volterra predator-prey model. We then showcased the KAN-ODE framework's ability to learn symbolic source terms and complete solution profiles in higher-complexity and data-lean scenarios including wave propagation and shock formation, the complex Schr\"odinger equation, and the Allen-Cahn phase separation equation. The successful training of KAN-ODEs, and their improved performance compared to traditional Neural ODEs, implies significant potential in leveraging this novel network architecture in myriad scientific machine learning applications for discovering hidden physics and predicting dynamic evolution.
Authors: Randy Davila
Abstract: Graph combinatorial optimization problems are widely applicable and notoriously difficult to compute; for example, consider the traveling salesman or facility location problems. In this paper, we explore the feasibility of using convolutional neural networks (CNNs) on graph images to predict the cardinality of combinatorial properties of random graphs and networks. Specifically, we use image representations of modified adjacency matrices of random graphs as training samples for a CNN model to predict the stability number of random graphs; where the stability number is the cardinality of a maximum set of vertices in a graph that contains no pairwise adjacency between vertices. The model and results presented in this study suggest potential for applying deep learning in combinatorial optimization problems previously not considered by simple deep learning techniques.
Authors: Federico Arangath Joseph, Kilian Konstantin Haefeli, Noah Liniger, Caglar Gulcehre
Abstract: This work explores the in-context learning capabilities of State Space Models (SSMs) and presents, to the best of our knowledge, the first theoretical explanation of a possible underlying mechanism. We introduce a novel weight construction for SSMs, enabling them to predict the next state of any dynamical system after observing previous states without parameter fine-tuning. This is accomplished by extending the HiPPO framework to demonstrate that continuous SSMs can approximate the derivative of any input signal. Specifically, we find an explicit weight construction for continuous SSMs and provide an asymptotic error bound on the derivative approximation. The discretization of this continuous SSM subsequently yields a discrete SSM that predicts the next state. Finally, we demonstrate the effectiveness of our parameterization empirically. This work should be an initial step toward understanding how sequence models based on SSMs learn in context.
Authors: Emine Akpinar, Sardar M. N. Islam, Murat Oduncuoglu
Abstract: The support vector machine algorithm with a quantum kernel estimator (QSVM-Kernel), as a leading example of a quantum machine learning technique, has undergone significant advancements. Nevertheless, its integration with classical data presents unique challenges. While quantum computers primarily interact with data in quantum states, embedding classical data into quantum states using feature mapping techniques is essential for leveraging quantum algorithms Despite the recognized importance of feature mapping, its specific impact on data classification outcomes remains largely unexplored. This study addresses this gap by comprehensively assessing the effects of various feature mapping methods on classification results, taking medical data analysis as a case study. In this study, the QSVM-Kernel method was applied to classification problems in two different and publicly available medical datasets, namely, the Wisconsin Breast Cancer (original) and The Cancer Genome Atlas (TCGA) Glioma datasets. In the QSVM-Kernel algorithm, quantum kernel matrices obtained from 9 different quantum feature maps were used. Thus, the effects of these quantum feature maps on the classification results of the QSVM-Kernel algorithm were examined in terms of both classifier performance and total execution time. As a result, in the Wisconsin Breast Cancer (original) and TCGA Glioma datasets, when Rx and Ry rotational gates were used, respectively, as feature maps in the QSVM-Kernel algorithm, the best classification performances were achieved both in terms of classification performance and total execution time. The contributions of this study are that (1) it highlights the significant impact of feature mapping techniques on medical data classification outcomes using the QSVM-Kernel algorithm, and (2) it also guides undertaking research for improved QSVM classification performance.
Authors: Yannis Karmim (CEDRIC - VERTIGO), Leshanshui Yang (LITIS), Rapha\"el Fournier S'Niehotta (CEDRIC - VERTIGO), Cl\'ement Chatelain (LITIS), S\'ebastien Adam (LITIS), Nicolas Thome (MLIA)
Abstract: Dynamic link prediction is a critical task in the analysis of evolving networks, with applications ranging from recommender systems to economic exchanges. However, the concept of the temporal receptive field, which refers to the temporal context that models use for making predictions, has been largely overlooked and insufficiently analyzed in existing research. In this study, we present a comprehensive analysis of the temporal receptive field in dynamic graph learning. By examining multiple datasets and models, we formalize the role of temporal receptive field and highlight their crucial influence on predictive accuracy. Our results demonstrate that appropriately chosen temporal receptive field can significantly enhance model performance, while for some models, overly large windows may introduce noise and reduce accuracy. We conduct extensive benchmarking to validate our findings, ensuring that all experiments are fully reproducible. Code is available at https://github.com/ykrmm/BenchmarkTW .
Authors: WooJae Jeon, KangJun Lee, Jeewoo Lee
Abstract: This paper introduces PG-Rainbow, a novel algorithm that incorporates a distributional reinforcement learning framework with a policy gradient algorithm. Existing policy gradient methods are sample inefficient and rely on the mean of returns when calculating the state-action value function, neglecting the distributional nature of returns in reinforcement learning tasks. To address this issue, we use an Implicit Quantile Network that provides the quantile information of the distribution of rewards to the critic network of the Proximal Policy Optimization algorithm. We show empirical results that through the integration of reward distribution information into the policy network, the policy agent acquires enhanced capabilities to comprehensively evaluate the consequences of potential actions in a given state, facilitating more sophisticated and informed decision-making processes. We evaluate the performance of the proposed algorithm in the Atari-2600 game suite, simulated via the Arcade Learning Environment (ALE).
Authors: Rui Zhu, Shuvrarghya Ghosh, Subhashis Ghosal
Abstract: Semi-supervised learning is a model training method that uses both labeled and unlabeled data. This paper proposes a fully Bayes semi-supervised learning algorithm that can be applied to any multi-category classification problem. We assume the labels are missing at random when using unlabeled data in a semi-supervised setting. Suppose we have $K$ classes in the data. We assume that the observations follow $K$ multivariate normal distributions depending on their true class labels after some common unknown transformation is applied to each component of the observation vector. The function is expanded in a B-splines series, and a prior is added to the coefficients. We consider a normal prior on the coefficients and constrain the values to meet the normality and identifiability constraints requirement. The precision matrices of the Gaussian distributions are given a conjugate Wishart prior, while the means are given the improper uniform prior. The resulting posterior is still conditionally conjugate, and the Gibbs sampler aided by a data-augmentation technique can thus be adopted. An extensive simulation study compares the proposed method with several other available methods. The proposed method is also applied to real datasets on diagnosing breast cancer and classification of signals. We conclude that the proposed method has a better prediction accuracy in various cases.
Authors: Hanrui Wang, Zhekai Zhang, Song Han
Abstract: The attention mechanism is becoming increasingly popular in Natural Language Processing (NLP) applications, showing superior performance than convolutional and recurrent architectures. However, attention becomes the compution bottleneck because of its quadratic computational complexity to input length, complicated data movement and low arithmetic intensity. Moreover, existing NN accelerators mainly focus on optimizing convolutional or recurrent models, and cannot efficiently support attention. In this paper, we present SpAtten, an efficient algorithm-architecture co-design that leverages token sparsity, head sparsity, and quantization opportunities to reduce the attention computation and memory access. Inspired by the high redundancy of human languages, we propose the novel cascade token pruning to prune away unimportant tokens in the sentence. We also propose cascade head pruning to remove unessential heads. Cascade pruning is fundamentally different from weight pruning since there is no trainable weight in the attention mechanism, and the pruned tokens and heads are selected on the fly. To efficiently support them on hardware, we design a novel top-k engine to rank token and head importance scores with high throughput. Furthermore, we propose progressive quantization that first fetches MSBs only and performs the computation; if the confidence is low, it fetches LSBs and recomputes the attention outputs, trading computation for memory reduction. Extensive experiments on 30 benchmarks show that, on average, SpAtten reduces DRAM access by 10.0x with no accuracy loss, and achieves 1.6x, 3.0x, 162x, 347x speedup, and 1,4x, 3.2x, 1193x, 4059x energy savings over A3 accelerator, MNNFast accelerator, TITAN Xp GPU, Xeon CPU, respectively.
Authors: Yirui Liu, Xinghao Qiao, Yulong Pei, Liying Wang
Abstract: This paper introduces the Deep Functional Factor Model (DF2M), a Bayesian nonparametric model designed for analysis of high-dimensional functional time series. DF2M is built upon the Indian Buffet Process and the multi-task Gaussian Process, incorporating a deep kernel function that captures non-Markovian and nonlinear temporal dynamics. Unlike many black-box deep learning models, DF2M offers an explainable approach to utilizing neural networks by constructing a factor model and integrating deep neural networks within the kernel function. Additionally, we develop a computationally efficient variational inference algorithm to infer DF2M. Empirical results from four real-world datasets demonstrate that DF2M provides better explainability and superior predictive accuracy compared to conventional deep learning models for high-dimensional functional time series.
Authors: Michael F. Liu, Saiyue Lyu, Margarita Vinaroz, Mijung Park
Abstract: Diffusion models (DMs) are one of the most widely used generative models for producing high quality images. However, a flurry of recent papers points out that DMs are least private forms of image generators, by extracting a significant number of near-identical replicas of training images from DMs. Existing privacy-enhancing techniques for DMs, unfortunately, do not provide a good privacy-utility tradeoff. In this paper, we aim to improve the current state of DMs with differential privacy (DP) by adopting the \textit{Latent} Diffusion Models (LDMs). LDMs are equipped with powerful pre-trained autoencoders that map the high-dimensional pixels into lower-dimensional latent representations, in which DMs are trained, yielding a more efficient and fast training of DMs. Rather than fine-tuning the entire LDMs, we fine-tune only the $\textit{attention}$ modules of LDMs with DP-SGD, reducing the number of trainable parameters by roughly $90\%$ and achieving a better privacy-accuracy trade-off. Our approach allows us to generate realistic, high-dimensional images (256x256) conditioned on text prompts with DP guarantees, which, to the best of our knowledge, has not been attempted before. Our approach provides a promising direction for training more powerful, yet training-efficient differentially private DMs, producing high-quality DP images. Our code is available at https://anonymous.4open.science/r/DP-LDM-4525.
Authors: Ziwei Fan, Zhiwei Liu, Hao Peng, Philip S. Yu
Abstract: Sequential recommendation (SR) models the dynamic user preferences and generates the next-item prediction as the affinity between the sequence and items, in a joint latent space with low dimensions (i.e., the sequence and item embedding space). Both sequence and item representations suffer from the representation degeneration issue due to the user/item long-tail distributions, where tail users/ items are indistinguishably distributed as a narrow cone in the latent space. We argue that the representation degeneration issue is the root cause of insufficient recommendation diversity in existing SR methods, impairing the user potential exploration and further worsening the echo chamber issue. In this work, we first disclose the connection between the representation degeneration and recommendation diversity, in which severer representation degeneration indicates lower recommendation diversity. We then propose a novel Singular sPectrum sMoothing regularization for Recommendation (SPMRec), which acts as a controllable surrogate to alleviate the degeneration and achieve the balance between recommendation diversity and performance. The proposed smoothing regularization alleviates the degeneration by maximizing the area under the singular value curve, which is also the diversity surrogate. We conduct experiments on four benchmark datasets to demonstrate the superiority of SPMRec, and show that the proposed singular spectrum smoothing can control the balance of recommendation performance and diversity simultaneously.
Authors: Kunal Talwar, Shan Wang, Audra McMillan, Vojta Jina, Vitaly Feldman, Pansy Bansal, Bailey Basile, Aine Cahill, Yi Sheng Chan, Mike Chatzidakis, Junye Chen, Oliver Chick, Mona Chitnis, Suman Ganta, Yusuf Goren, Filip Granqvist, Kristine Guo, Frederic Jacobs, Omid Javidbakht, Albert Liu, Richard Low, Dan Mascenik, Steve Myers, David Park, Wonhee Park, Gianni Parsa, Tommy Pauly, Christian Priebe, Rehan Rishi, Guy Rothblum, Michael Scaria, Linmao Song, Congzheng Song, Karl Tarbe, Sebastian Vogt, Luke Winstrom, Shundong Zhou
Abstract: We revisit the problem of designing scalable protocols for private statistics and private federated learning when each device holds its private data. Locally differentially private algorithms require little trust but are (provably) limited in their utility. Centrally differentially private algorithms can allow significantly better utility but require a trusted curator. This gap has led to significant interest in the design and implementation of simple cryptographic primitives, that can allow central-like utility guarantees without having to trust a central server. Our first contribution is to propose a new primitive that allows for efficient implementation of several commonly used algorithms, and allows for privacy accounting that is close to that in the central setting without requiring the strong trust assumptions it entails. {\em Shuffling} and {\em aggregation} primitives that have been proposed in earlier works enable this for some algorithms, but have significant limitations as primitives. We propose a {\em Samplable Anonymous Aggregation} primitive, which computes an aggregate over a random subset of the inputs and show that it leads to better privacy-utility trade-offs for various fundamental tasks. Secondly, we propose a system architecture that implements this primitive and perform a security analysis of the proposed system. Our design combines additive secret-sharing with anonymization and authentication infrastructures.
Authors: Tobias Christian Nauen, Sebastian Palacio, Federico Raue, Andreas Dengel
Abstract: Self-attention in Transformers comes with a high computational cost because of their quadratic computational complexity, but their effectiveness in addressing problems in language and vision has sparked extensive research aimed at enhancing their efficiency. However, diverse experimental conditions, spanning multiple input domains, prevent a fair comparison based solely on reported results, posing challenges for model selection. To address this gap in comparability, we perform a large-scale benchmark of more than 45 models for image classification, evaluating key efficiency aspects, including accuracy, speed, and memory usage. Our benchmark provides a standardized baseline for efficiency-oriented transformers. We analyze the results based on the Pareto front -- the boundary of optimal models. Surprisingly, despite claims of other models being more efficient, ViT remains Pareto optimal across multiple metrics. We observe that hybrid attention-CNN models exhibit remarkable inference memory- and parameter-efficiency. Moreover, our benchmark shows that using a larger model in general is more efficient than using higher resolution images. Thanks to our holistic evaluation, we provide a centralized resource for practitioners and researchers, facilitating informed decisions when selecting or developing efficient transformers.
Authors: Dahun Kim, Anelia Angelova, Weicheng Kuo
Abstract: We present a new open-vocabulary detection approach based on region-centric image-language pretraining to bridge the gap between image-level pretraining and open-vocabulary object detection. At the pretraining phase, we incorporate the detector architecture on top of the classification backbone, which better serves the region-level recognition needs of detection by enabling the detector heads to learn from large-scale image-text pairs. Using only standard contrastive loss and no pseudo-labeling, our approach is a simple yet effective extension of the contrastive learning method to learn emergent object-semantic cues. In addition, we propose a shifted-window learning approach upon window attention to make the backbone representation more robust, translation-invariant, and less biased by the window pattern. On the popular LVIS open-vocabulary detection benchmark, our approach sets a new state of the art of 37.6 mask APr using the common ViT-L backbone and public LAION dataset, and 40.5 mask APr using the DataComp-1B dataset, significantly outperforming the best existing approach by +3.7 mask APr at system level. On the COCO benchmark, we achieve very competitive 39.6 novel AP without pseudo labeling or weak supervision. In addition, we evaluate our approach on the transfer detection setup, where it demonstrates notable improvement over the baseline. Visualization reveals emerging object locality from the pretraining recipes compared to the baseline.
Authors: Sandesh Pokhrel, Sanjay Bhandari, Eduard Vazquez, Yash Raj Shrestha, Binod Bhattarai
Abstract: Coronary Artery Diseases (CADs) although preventable, are one of the leading causes of death and disability. Diagnosis of these diseases is often difficult and resource intensive. Angiographic imaging segmentation of the arteries has evolved as a tool of assistance that helps clinicians make an accurate diagnosis. However, due to the limited amount of data and the difficulty in curating a dataset, the task of segmentation has proven challenging. In this study, we introduce the use of pseudo-labels to address the issue of limited data in the angiographic dataset to enhance the performance of the baseline YOLO model. Unlike existing data augmentation techniques that improve the model constrained to a fixed dataset, we introduce the use of pseudo-labels generated on a dataset of separate related task to diversify and improve model performance. This method increases the baseline F1 score by 9% in the validation data set and by 3% in the test data set.
Authors: Haizhong Zheng, Jiachen Sun, Shutong Wu, Bhavya Kailkhura, Zhuoqing Mao, Chaowei Xiao, Atul Prakash
Abstract: Given a real-world dataset, data condensation (DC) aims to synthesize a small synthetic dataset that captures the knowledge of a natural dataset while being usable for training models with comparable accuracy. Recent works propose to enhance DC with data parameterization, which condenses data into very compact parameterized data containers instead of images. The intuition behind data parameterization is to encode shared features of images to avoid additional storage costs. In this paper, we recognize that images share common features in a hierarchical way due to the inherent hierarchical structure of the classification system, which is overlooked by current data parameterization methods. To better align DC with this hierarchical nature and encourage more efficient information sharing inside data containers, we propose a novel data parameterization architecture, Hierarchical Memory Network (HMN). HMN stores condensed data in a three-tier structure, representing the dataset-level, class-level, and instance-level features. Another helpful property of the hierarchical architecture is that HMN naturally ensures good independence among images despite achieving information sharing. This enables instance-level pruning for HMN to reduce redundant information, thereby further minimizing redundancy and enhancing performance. We evaluate HMN on five public datasets and show that our proposed method outperforms all baselines.
Authors: Zheyu Oliver Wang, Ricardo Baptista, Youssef Marzouk, Lars Ruthotto, Deepanshu Verma
Abstract: We present two neural network approaches that approximate the solutions of static and dynamic conditional optimal transport (COT) problems. Both approaches enable conditional sampling and conditional density estimation, which are core tasks in Bayesian inference$\unicode{x2013}$particularly in the simulation-based ("likelihood-free") setting. Our methods represent the target conditional distributions as transformations of a tractable reference distribution and, therefore, fall into the framework of measure transport. Although many measure transport approaches model the transformation as COT maps, obtaining the map is computationally challenging, even in moderate dimensions. To improve scalability, our numerical algorithms use neural networks to parameterize COT maps and further exploit the structure of the COT problem. Our static approach approximates the map as the gradient of a partially input-convex neural network. It uses a novel numerical implementation to increase computational efficiency compared to state-of-the-art alternatives. Our dynamic approach approximates the conditional optimal transport via the flow map of a regularized neural ODE; compared to the static approach, it is slower to train but offers more modeling choices and can lead to faster sampling. We demonstrate both algorithms numerically, comparing them with competing state-of-the-art approaches, using benchmark datasets and simulation-based Bayesian inverse problems.
Authors: Jiakai Wang, Xianglong Liu, Jin Hu, Donghua Wang, Siyang Wu, Tingsong Jiang, Yuanfang Guo, Aishan Liu, Aishan Liu, Jiantao Zhou
Abstract: Deep neural networks (DNNs) have demonstrated high vulnerability to adversarial examples, raising broad security concerns about their applications. Besides the attacks in the digital world, the practical implications of adversarial examples in the physical world present significant challenges and safety concerns. However, current research on physical adversarial examples (PAEs) lacks a comprehensive understanding of their unique characteristics, leading to limited significance and understanding. In this paper, we address this gap by thoroughly examining the characteristics of PAEs within a practical workflow encompassing training, manufacturing, and re-sampling processes. By analyzing the links between physical adversarial attacks, we identify manufacturing and re-sampling as the primary sources of distinct attributes and particularities in PAEs. Leveraging this knowledge, we develop a comprehensive analysis and classification framework for PAEs based on their specific characteristics, covering over 100 studies on physical-world adversarial examples. Furthermore, we investigate defense strategies against PAEs and identify open challenges and opportunities for future research. We aim to provide a fresh, thorough, and systematic understanding of PAEs, thereby promoting the development of robust adversarial learning and its application in open-world scenarios to provide the community with a continuously updated list of physical world adversarial sample resources, including papers, code, \etc, within the proposed framework
Authors: Michael Lan, Philip Torr, Fazl Barez
Abstract: While transformer models exhibit strong capabilities on linguistic tasks, their complex architectures make them difficult to interpret. Recent work has aimed to reverse engineer transformer models into human-readable representations called circuits that implement algorithmic functions. We extend this research by analyzing and comparing circuits for similar sequence continuation tasks, which include increasing sequences of Arabic numerals, number words, and months. By applying circuit interpretability analysis, we identify a key sub-circuit in both GPT-2 Small and Llama-2-7B responsible for detecting sequence members and for predicting the next member in a sequence. Our analysis reveals that semantically related sequences rely on shared circuit subgraphs with analogous roles. Additionally, we show that this sub-circuit has effects on various math-related prompts, such as on intervaled circuits, Spanish number word and months continuation, and natural language word problems. Overall, documenting shared computational structures enables better model behavior predictions, identification of errors, and safer editing procedures. This mechanistic understanding of transformers is a critical step towards building more robust, aligned, and interpretable language models.
Authors: Dylan Green, Stephen Bailey
Abstract: Non-negative matrix factorization (NMF) is a dimensionality reduction technique that has shown promise for analyzing noisy data, especially astronomical data. For these datasets, the observed data may contain negative values due to noise even when the true underlying physical signal is strictly positive. Prior NMF work has not treated negative data in a statistically consistent manner, which becomes problematic for low signal-to-noise data with many negative values. In this paper we present two algorithms, Shift-NMF and Nearly-NMF, that can handle both the noisiness of the input data and also any introduced negativity. Both of these algorithms use the negative data space without clipping, and correctly recover non-negative signals without any introduced positive offset that occurs when clipping negative data. We demonstrate this numerically on both simple and more realistic examples, and prove that both algorithms have monotonically decreasing update rules.
Authors: Stefano Bann\`o, Rao Ma, Mengjie Qian, Kate M. Knill, Mark J. F. Gales
Abstract: Grammatical feedback is crucial for L2 learners, teachers, and testers. Spoken grammatical error correction (GEC) aims to supply feedback to L2 learners on their use of grammar when speaking. This process usually relies on a cascaded pipeline comprising an ASR system, disfluency removal, and GEC, with the associated concern of propagating errors between these individual modules. In this paper, we introduce an alternative "end-to-end" approach to spoken GEC, exploiting a speech recognition foundation model, Whisper. This foundation model can be used to replace the whole framework or part of it, e.g., ASR and disfluency removal. These end-to-end approaches are compared to more standard cascaded approaches on the data obtained from a free-speaking spoken language assessment test, Linguaskill. Results demonstrate that end-to-end spoken GEC is possible within this architecture, but the lack of available data limits current performance compared to a system using large quantities of text-based GEC data. Conversely, end-to-end disfluency detection and removal, which is easier for the attention-based Whisper to learn, does outperform cascaded approaches. Additionally, the paper discusses the challenges of providing feedback to candidates when using end-to-end systems for spoken GEC.
Authors: Nikhil Parthasarathy, Olivier J. H\'enaff, Eero P. Simoncelli
Abstract: Human ability to recognize complex visual patterns arises through transformations performed by successive areas in the ventral visual cortex. Deep neural networks trained end-to-end for object recognition approach human capabilities, and offer the best descriptions to date of neural responses in the late stages of the hierarchy. But these networks provide a poor account of the early stages, compared to traditional hand-engineered models, or models optimized for coding efficiency or prediction. Moreover, the gradient backpropagation used in end-to-end learning is generally considered to be biologically implausible. Here, we overcome both of these limitations by developing a bottom-up self-supervised training methodology that operates independently on successive layers. Specifically, we maximize feature similarity between pairs of locally-deformed natural image patches, while decorrelating features across patches sampled from other images. Crucially, the deformation amplitudes are adjusted proportionally to receptive field sizes in each layer, thus matching the task complexity to the capacity at each stage of processing. In comparison with architecture-matched versions of previous models, we demonstrate that our layerwise complexity-matched learning (LCL) formulation produces a two-stage model (LCL-V2) that is better aligned with selectivity properties and neural activity in primate area V2. We demonstrate that the complexity-matched learning paradigm is responsible for much of the emergence of the improved biological alignment. Finally, when the two-stage model is used as a fixed front-end for a deep network trained to perform object recognition, the resultant model (LCL-V2Net) is significantly better than standard end-to-end self-supervised, supervised, and adversarially-trained models in terms of generalization to out-of-distribution tasks and alignment with human behavior.
Authors: Weijia Zhang, Chun Kai Ling, Xuanhui Zhang
Abstract: Censoring is the central problem in survival analysis where either the time-to-event (for instance, death), or the time-tocensoring (such as loss of follow-up) is observed for each sample. The majority of existing machine learning-based survival analysis methods assume that survival is conditionally independent of censoring given a set of covariates; an assumption that cannot be verified since only marginal distributions is available from the data. The existence of dependent censoring, along with the inherent bias in current estimators has been demonstrated in a variety of applications, accentuating the need for a more nuanced approach. However, existing methods that adjust for dependent censoring require practitioners to specify the ground truth copula. This requirement poses a significant challenge for practical applications, as model misspecification can lead to substantial bias. In this work, we propose a flexible deep learning-based survival analysis method that simultaneously accommodate for dependent censoring and eliminates the requirement for specifying the ground truth copula. We theoretically prove the identifiability of our model under a broad family of copulas and survival distributions. Experiments results from a wide range of datasets demonstrate that our approach successfully discerns the underlying dependency structure and significantly reduces survival estimation bias when compared to existing methods.
Authors: Jason Yim, Andrew Campbell, Emile Mathieu, Andrew Y. K. Foong, Michael Gastegger, Jos\'e Jim\'enez-Luna, Sarah Lewis, Victor Garcia Satorras, Bastiaan S. Veeling, Frank No\'e, Regina Barzilay, Tommi S. Jaakkola
Abstract: Protein design often begins with the knowledge of a desired function from a motif which motif-scaffolding aims to construct a functional protein around. Recently, generative models have achieved breakthrough success in designing scaffolds for a range of motifs. However, generated scaffolds tend to lack structural diversity, which can hinder success in wet-lab validation. In this work, we extend FrameFlow, an SE(3) flow matching model for protein backbone generation, to perform motif-scaffolding with two complementary approaches. The first is motif amortization, in which FrameFlow is trained with the motif as input using a data augmentation strategy. The second is motif guidance, which performs scaffolding using an estimate of the conditional score from FrameFlow without additional training. On a benchmark of 24 biologically meaningful motifs, we show our method achieves 2.5 times more designable and unique motif-scaffolds compared to state-of-the-art. Code: https://github.com/microsoft/protein-frame-flow
Authors: Nikita Soni, Niranjan Balasubramanian, H. Andrew Schwartz, Dirk Hovy
Abstract: Pre-trained language models consider the context of neighboring words and documents but lack any author context of the human generating the text. However, language depends on the author's states, traits, social, situational, and environmental attributes, collectively referred to as human context (Soni et al., 2024). Human-centered natural language processing requires incorporating human context into language models. Currently, two methods exist: pre-training with 1) group-wise attributes (e.g., over-45-year-olds) or 2) individual traits. Group attributes are simple but coarse -- not all 45-year-olds write the same way -- while individual traits allow for more personalized representations, but require more complex modeling and data. It is unclear which approach benefits what tasks. We compare pre-training models with human context via 1) group attributes, 2) individual users, and 3) a combined approach on five user- and document-level tasks. Our results show that there is no best approach, but that human-centered language modeling holds avenues for different methods.
Authors: Soojung Yang, Juno Nam, Johannes C. B. Dietschreit, Rafael G\'omez-Bombarelli
Abstract: In molecular dynamics simulations, rare events, such as protein folding, are typically studied using enhanced sampling techniques, most of which are based on the definition of a collective variable (CV) along which acceleration occurs. Obtaining an expressive CV is crucial, but often hindered by the lack of information about the particular event, e.g., the transition from unfolded to folded conformation. We propose a simulation-free data augmentation strategy using physics-inspired metrics to generate geodesic interpolations resembling protein folding transitions, thereby improving sampling efficiency without true transition state samples. This new data can be used to improve the accuracy of classifier-based methods. Alternatively, a regression-based learning scheme for CV models can be adopted by leveraging the interpolation progress parameter.
Authors: Peter Orbanz
Abstract: Consider a convex function that is invariant under an group of transformations. If it has a minimizer, does it also have an invariant minimizer? Variants of this problem appear in nonparametric statistics and in a number of adjacent fields. The answer depends on the choice of function, and on what one may loosely call the geometry of the problem -- the interplay between convexity, the group, and the underlying vector space, which is typically infinite-dimensional. We observe that this geometry is completely encoded in the smallest closed convex invariant subsets of the space, and proceed to study these sets, for groups that are amenable but not necessarily compact. We then apply this toolkit to the invariant optimality problem. It yields new results on invariant kernel mean embeddings and risk-optimal invariant couplings, and clarifies relations between seemingly distinct ideas, such as the summation trick used in machine learning to construct equivariant neural networks and the classic Hunt-Stein theorem of statistics.
Authors: Jiwon Song, Kyungseok Oh, Taesu Kim, Hyungjun Kim, Yulhwa Kim, Jae-Joon Kim
Abstract: Large language models (LLMs) have proven to be highly effective across various natural language processing tasks. However, their large number of parameters poses significant challenges for practical deployment. Pruning, a technique aimed at reducing the size and complexity of LLMs, offers a potential solution by removing redundant components from the network. Despite the promise of pruning, existing methods often struggle to achieve substantial end-to-end LLM inference speedup. In this paper, we introduce SLEB, a novel approach designed to streamline LLMs by eliminating redundant transformer blocks. We choose the transformer block as the fundamental unit for pruning, because LLMs exhibit block-level redundancy with high similarity between the outputs of neighboring blocks. This choice allows us to effectively enhance the processing speed of LLMs. Our experimental results demonstrate that SLEB outperforms previous LLM pruning methods in accelerating LLM inference while also maintaining superior perplexity and accuracy, making SLEB as a promising technique for enhancing the efficiency of LLMs. The code is available at: https://github.com/jiwonsong-dev/SLEB.
Authors: Yingru Li
Abstract: We present a simplified and unified analysis of the Johnson-Lindenstrauss (JL) lemma, a cornerstone of dimensionality reduction for managing high-dimensional data. Our approach simplifies understanding and unifies various constructions under the JL framework, including spherical, binary-coin, sparse JL, Gaussian, and sub-Gaussian models. This unification preserves the intrinsic geometry of data, essential for applications from streaming algorithms to reinforcement learning. We provide the first rigorous proof of the spherical construction's effectiveness and introduce a general class of sub-Gaussian constructions within this simplified framework. Central to our contribution is an innovative extension of the Hanson-Wright inequality to high dimensions, complete with explicit constants. By using simple yet powerful probabilistic tools and analytical techniques, such as an enhanced diagonalization process, our analysis solidifies the theoretical foundation of the JL lemma by removing an independence assumption and extends its practical applicability to contemporary algorithms.
Authors: Junlin Han, Filippos Kokkinos, Philip Torr
Abstract: This paper presents a novel method for building scalable 3D generative models utilizing pre-trained video diffusion models. The primary obstacle in developing foundation 3D generative models is the limited availability of 3D data. Unlike images, texts, or videos, 3D data are not readily accessible and are difficult to acquire. This results in a significant disparity in scale compared to the vast quantities of other types of data. To address this issue, we propose using a video diffusion model, trained with extensive volumes of text, images, and videos, as a knowledge source for 3D data. By unlocking its multi-view generative capabilities through fine-tuning, we generate a large-scale synthetic multi-view dataset to train a feed-forward 3D generative model. The proposed model, VFusion3D, trained on nearly 3M synthetic multi-view data, can generate a 3D asset from a single image in seconds and achieves superior performance when compared to current SOTA feed-forward 3D generative models, with users preferring our results over 90% of the time.
Authors: Tim Salzmann, Markus Ryll, Alex Bewley, Matthias Minderer
Abstract: Visual relationship detection aims to identify objects and their relationships in images. Prior methods approach this task by adding separate relationship modules or decoders to existing object detection architectures. This separation increases complexity and hinders end-to-end training, which limits performance. We propose a simple and highly efficient decoder-free architecture for open-vocabulary visual relationship detection. Our model consists of a Transformer-based image encoder that represents objects as tokens and models their relationships implicitly. To extract relationship information, we introduce an attention mechanism that selects object pairs likely to form a relationship. We provide a single-stage recipe to train this model on a mixture of object and relationship detection data. Our approach achieves state-of-the-art relationship detection performance on Visual Genome and on the large-vocabulary GQA benchmark at real-time inference speeds. We provide ablations, real-world qualitative examples, and analyses of zero-shot performance.
Authors: Thomas Merth, Qichen Fu, Mohammad Rastegari, Mahyar Najibi
Abstract: Despite the successes of large language models (LLMs), they exhibit significant drawbacks, particularly when processing long contexts. Their inference cost scales quadratically with respect to sequence length, making it expensive for deployment in some real-world text processing applications, such as retrieval-augmented generation (RAG). Additionally, LLMs also exhibit the "distraction phenomenon", where irrelevant context in the prompt degrades output quality. To address these drawbacks, we propose a novel RAG prompting methodology, *superposition prompting*, which can be directly applied to pre-trained transformer-based LLMs *without the need for fine-tuning*. At a high level, superposition prompting allows the LLM to process input documents in parallel *prompt paths*, discarding paths once they are deemed irrelevant. We demonstrate the capability of our method to simultaneously enhance time efficiency across a variety of question-answering benchmarks using multiple pre-trained LLMs. Furthermore, our technique significantly improves accuracy when the retrieved context is large relative the context the model was trained on. For example, our approach facilitates a 93x reduction in compute time while *improving* accuracy by 43% on the NaturalQuestions-Open dataset with the MPT-7B instruction-tuned model over naive RAG.
Authors: Wanting Yang, Zehui Xiong, Yanli Yuan, Wenchao Jiang, Tony Q. S. Quek, Merouane Debbah
Abstract: In the era of 6G, with compelling visions of intelligent transportation systems and digital twins, remote surveillance is poised to become a ubiquitous practice. Substantial data volume and frequent updates present challenges in wireless networks. To address these challenges, we propose a novel agent-driven generative semantic communication (A-GSC) framework based on reinforcement learning. In contrast to the existing research on semantic communication (SemCom), which mainly focuses on either semantic extraction or semantic sampling, we seamlessly integrate both by jointly considering the intrinsic attributes of source information and the contextual information regarding the task. Notably, the introduction of generative artificial intelligence (GAI) enables the independent design of semantic encoders and decoders. In this work, we develop an agent-assisted semantic encoder with cross-modality capability, which can track the semantic changes, channel condition, to perform adaptive semantic extraction and sampling. Accordingly, we design a semantic decoder with both predictive and generative capabilities, consisting of two tailored modules. Moreover, the effectiveness of the designed models has been verified using the UA-DETRAC dataset, demonstrating the performance gains of the overall A-GSC framework in both energy saving and reconstruction accuracy.
Authors: Nima Shahbazi, Mahdi Erfanian, Abolfazl Asudeh, Fatemeh Nargesian, Divesh Srivastava
Abstract: Entity matching is one the earliest tasks that occur in the big data pipeline and is alarmingly exposed to unintentional biases that affect the quality of data. Identifying and mitigating the biases that exist in the data or are introduced by the matcher at this stage can contribute to promoting fairness in downstream tasks. This demonstration showcases FairEM360, a framework for 1) auditing the output of entity matchers across a wide range of fairness measures and paradigms, 2) providing potential explanations for the underlying reasons for unfairness, and 3) providing resolutions for the unfairness issues through an exploratory process with human-in-the-loop feedback, utilizing an ensemble of matchers. We aspire for FairEM360 to contribute to the prioritization of fairness as a key consideration in the evaluation of EM pipelines.
Authors: Charig Yang, Weidi Xie, Andrew Zisserman
Abstract: Our objective is to discover and localize monotonic temporal changes in a sequence of images. To achieve this, we exploit a simple proxy task of ordering a shuffled image sequence, with `time' serving as a supervisory signal, since only changes that are monotonic with time can give rise to the correct ordering. We also introduce a transformer-based model for ordering of image sequences of arbitrary length with built-in attribution maps. After training, the model successfully discovers and localizes monotonic changes while ignoring cyclic and stochastic ones. We demonstrate applications of the model in multiple domains covering different scene and object types, discovering both object-level and environmental changes in unseen sequences. We also demonstrate that the attention-based attribution maps function as effective prompts for segmenting the changing regions, and that the learned representations can be used for downstream applications. Finally, we show that the model achieves the state-of-the-art on standard benchmarks for image ordering.
Authors: Joseph A. Vincent, Haruki Nishimura, Masha Itkina, Paarth Shah, Mac Schwager, Thomas Kollar
Abstract: With the rise of stochastic generative models in robot policy learning, end-to-end visuomotor policies are increasingly successful at solving complex tasks by learning from human demonstrations. Nevertheless, since real-world evaluation costs afford users only a small number of policy rollouts, it remains a challenge to accurately gauge the performance of such policies. This is exacerbated by distribution shifts causing unpredictable changes in performance during deployment. To rigorously evaluate behavior cloning policies, we present a framework that provides a tight lower-bound on robot performance in an arbitrary environment, using a minimal number of experimental policy rollouts. Notably, by applying the standard stochastic ordering to robot performance distributions, we provide a worst-case bound on the entire distribution of performance (via bounds on the cumulative distribution function) for a given task. We build upon established statistical results to ensure that the bounds hold with a user-specified confidence level and tightness, and are constructed from as few policy rollouts as possible. In experiments we evaluate policies for visuomotor manipulation in both simulation and hardware. Specifically, we (i) empirically validate the guarantees of the bounds in simulated manipulation settings, (ii) find the degree to which a learned policy deployed on hardware generalizes to new real-world environments, and (iii) rigorously compare two policies tested in out-of-distribution settings. Our experimental data, code, and implementation of confidence bounds are open-source.
Authors: Siva Uday Sampreeth Chebolu, Franck Dernoncourt, Nedim Lipka, Thamar Solorio
Abstract: Aspect-Based Sentiment Analysis (ABSA) has experienced tremendous expansion and diversity due to various shared tasks spanning several languages and fields and organized via SemEval workshops and Germeval. Nonetheless, a few shortcomings still need to be addressed, such as the lack of low-resource language evaluations and the emphasis on sentence-level analysis. To thoroughly assess ABSA techniques in the context of complete reviews, this research presents a novel task, Review-Level Opinion Aspect Sentiment Target (ROAST). ROAST seeks to close the gap between sentence-level and text-level ABSA by identifying every ABSA constituent at the review level. We extend the available datasets to enable ROAST, addressing the drawbacks noted in previous research by incorporating low-resource languages, numerous languages, and a variety of topics. Through this effort, ABSA research will be able to cover more ground and get a deeper comprehension of the task and its practical application in a variety of languages and domains (https://github.com/RiTUAL-UH/ROAST-ABSA).
Authors: Sushant Agarwal, Gautam Kamath, Mahbod Majid, Argyris Mouzakis, Rose Silver, Jonathan Ullman
Abstract: We study person-level differentially private (DP) mean estimation in the case where each person holds multiple samples. DP here requires the usual notion of distributional stability when $\textit{all}$ of a person's datapoints can be modified. Informally, if $n$ people each have $m$ samples from an unknown $d$-dimensional distribution with bounded $k$-th moments, we show that \[n = \tilde \Theta\left(\frac{d}{\alpha^2 m} + \frac{d}{\alpha m^{1/2} \varepsilon} + \frac{d}{\alpha^{k/(k-1)} m \varepsilon} + \frac{d}{\varepsilon}\right)\] people are necessary and sufficient to estimate the mean up to distance $\alpha$ in $\ell_2$-norm under $\varepsilon$-differential privacy (and its common relaxations). In the multivariate setting, we give computationally efficient algorithms under approximate-DP and computationally inefficient algorithms under pure DP, and our nearly matching lower bounds hold for the most permissive case of approximate DP. Our computationally efficient estimators are based on the standard clip-and-noise framework, but the analysis for our setting requires both new algorithmic techniques and new analyses. In particular, our new bounds on the tails of sums of independent, vector-valued, bounded-moments random variables may be of interest.
Authors: Ruizhe Huang, Xiaohui Zhang, Zhaoheng Ni, Li Sun, Moto Hira, Jeff Hwang, Vimal Manohar, Vineel Pratap, Matthew Wiesner, Shinji Watanabe, Daniel Povey, Sanjeev Khudanpur
Abstract: Connectionist temporal classification (CTC) models are known to have peaky output distributions. Such behavior is not a problem for automatic speech recognition (ASR), but it can cause inaccurate forced alignments (FA), especially at finer granularity, e.g., phoneme level. This paper aims at alleviating the peaky behavior for CTC and improve its suitability for forced alignment generation, by leveraging label priors, so that the scores of alignment paths containing fewer blanks are boosted and maximized during training. As a result, our CTC model produces less peaky posteriors and is able to more accurately predict the offset of the tokens besides their onset. It outperforms the standard CTC model and a heuristics-based approach for obtaining CTC's token offset timestamps by 12-40% in phoneme and word boundary errors (PBE and WBE) measured on the Buckeye and TIMIT data. Compared with the most widely used FA toolkit Montreal Forced Aligner (MFA), our method performs similarly on PBE/WBE on Buckeye, yet falls behind MFA on TIMIT. Nevertheless, our method has a much simpler training pipeline and better runtime efficiency. Our training recipe and pretrained model are released in TorchAudio.
Authors: Jianuo Huang
Abstract: In recent advancements in Multi-agent Reinforcement Learning (MARL), its application has extended to various safety-critical scenarios. However, most methods focus on online learning, which presents substantial risks when deployed in real-world settings. Addressing this challenge, we introduce an innovative framework integrating diffusion models within the MARL paradigm. This approach notably enhances the safety of actions taken by multiple agents through risk mitigation while modeling coordinated action. Our framework is grounded in the Centralized Training with Decentralized Execution (CTDE) architecture, augmented by a Diffusion Model for prediction trajectory generation. Additionally, we incorporate a specialized algorithm to further ensure operational safety. We evaluate our model against baselines on the DSRL benchmark. Experiment results demonstrate that our model not only adheres to stringent safety constraints but also achieves superior performance compared to existing methodologies. This underscores the potential of our approach in advancing the safety and efficacy of MARL in real-world applications.
Authors: Lu\'isa Shimabucoro, Sebastian Ruder, Julia Kreutzer, Marzieh Fadaee, Sara Hooker
Abstract: The widespread adoption of synthetic data raises new questions about how models generating the data can influence other large language models (LLMs) via distilled data. To start, our work exhaustively characterizes the impact of passive inheritance of model properties by systematically studying the consequences of synthetic data integration. We provide one of the most comprehensive studies to-date of how the source of synthetic data shapes models' internal biases, calibration and generations' textual attributes and preferences. We find that models are surprisingly sensitive towards certain attributes even when the synthetic data prompts appear "neutral". which invites the question whether this sensitivity can be exploited for good. Our findings invite the question can we explicitly steer the models towards the properties we want at test time by exploiting the data generation process? This would have historically been considered infeasible due to the cost of collecting data with a specific characteristic or objective in mind. However, improvement in the quality of synthetic data, as well as a shift towards general-purpose models designed to follow a diverse way of instructions, means this question is timely. We propose active inheritance as a term to describe intentionally constraining synthetic data according to a non-differentiable objective. We demonstrate how active inheritance can steer the generation profiles of models towards desirable non-differentiable attributes, e.g. high lexical diversity or low toxicity.
Authors: Mengjie Qian, Siyuan Tang, Rao Ma, Kate M. Knill, Mark J. F. Gales
Abstract: Foundation ASR models often support many languages, e.g. 100 languages in Whisper. However, there has been limited work on integrating an additional, typically low-resource, language, while maintaining performance on the original language set. Fine-tuning, while simple, may degrade the accuracy of the original set. We compare three approaches that exploit adaptation parameters: soft language code tuning, train only the language code; soft prompt tuning, train prepended tokens; and LoRA where a small set of additional parameters are optimised. Elastic Weight Consolidation (EWC) offers an alternative compromise with the potential to maintain performance in specific target languages. Results show that direct fine-tuning yields the best performance for the new language but degrades existing language capabilities. EWC can address this issue for specific languages. If only adaptation parameters are used, the language capabilities are maintained but at the cost of performance in the new language.
Authors: Igor Sterner, Shiye Su, Petar Veli\v{c}kovi\'c
Abstract: We explore graph rewiring methods that optimise commute time. Recent graph rewiring approaches facilitate long-range interactions in sparse graphs, making such rewirings commute-time-optimal $\textit{on average}$. However, when an expert prior exists on which node pairs should or should not interact, a superior rewiring would favour short commute times between these privileged node pairs. We construct two synthetic datasets with known priors reflecting realistic settings, and use these to motivate two bespoke rewiring methods that incorporate the known prior. We investigate the regimes where our rewiring improves test performance on the synthetic datasets. Finally, we perform a case study on a real-world citation graph to investigate the practical implications of our work.
Authors: Xianing Chen, Hanting Chen, Hailin Hu
Abstract: Frequency domain representation of time series feature offers a concise representation for handling real-world time series data with inherent complexity and dynamic nature. However, current frequency-based methods with complex operations still fall short of state-of-the-art time domain methods for general time series analysis. In this work, we present Omni-Dimensional Frequency Learner (ODFL) model based on a in depth analysis among all the three aspects of the spectrum feature: channel redundancy property among the frequency dimension, the sparse and un-salient frequency energy distribution among the frequency dimension, and the semantic diversity among the variable dimension. Technically, our method is composed of a semantic-adaptive global filter with attention to the un-salient frequency bands and partial operation among the channel dimension. Empirical results show that ODFL achieves consistent state-of-the-art in five mainstream time series analysis tasks, including short- and long-term forecasting, imputation, classification, and anomaly detection, offering a promising foundation for time series analysis.
Authors: Maksym Andriushchenko, Nicolas Flammarion
Abstract: Refusal training is widely used to prevent LLMs from generating harmful, undesirable, or illegal outputs. We reveal a curious generalization gap in the current refusal training approaches: simply reformulating a harmful request in the past tense (e.g., "How to make a Molotov cocktail?" to "How did people make a Molotov cocktail?") is often sufficient to jailbreak many state-of-the-art LLMs. We systematically evaluate this method on Llama-3 8B, Claude-3.5 Sonnet, GPT-3.5 Turbo, Gemma-2 9B, Phi-3-Mini, GPT-4o mini, GPT-4o, and R2D2 models using GPT-3.5 Turbo as a reformulation model. For example, the success rate of this simple attack on GPT-4o increases from 1% using direct requests to 88% using 20 past tense reformulation attempts on harmful requests from JailbreakBench with GPT-4 as a jailbreak judge. Interestingly, we also find that reformulations in the future tense are less effective, suggesting that refusal guardrails tend to consider past historical questions more benign than hypothetical future questions. Moreover, our experiments on fine-tuning GPT-3.5 Turbo show that defending against past reformulations is feasible when past tense examples are explicitly included in the fine-tuning data. Overall, our findings highlight that the widely used alignment techniques -- such as SFT, RLHF, and adversarial training -- employed to align the studied models can be brittle and do not always generalize as intended. We provide code and jailbreak artifacts at https://github.com/tml-epfl/llm-past-tense.
Authors: Aditya Makkar, Divya Makkar, Aarav Patel, Liam Hebert
Abstract: The lack of fluency in sign language remains a barrier to seamless communication for hearing and speech-impaired communities. In this work, we propose a low-cost, real-time ASL-to-speech translation glove and an exhaustive training dataset of sign language patterns. We then benchmarked this dataset with supervised learning models, such as LSTMs, GRUs and Transformers, where our best model achieved 92% accuracy. The SignSpeak dataset has 7200 samples encompassing 36 classes (A-Z, 1-10) and aims to capture realistic signing patterns by using five low-cost flex sensors to measure finger positions at each time step at 36 Hz. Our open-source dataset, models and glove designs, provide an accurate and efficient ASL translator while maintaining cost-effectiveness, establishing a framework for future work to build on.
Authors: Irina Jurenka, Markus Kunesch, Kevin R. McKee, Daniel Gillick, Shaojian Zhu, Sara Wiltberger, Shubham Milind Phal, Katherine Hermann, Daniel Kasenberg, Avishkar Bhoopchand, Ankit Anand, Miruna P\^islar, Stephanie Chan, Lisa Wang, Jennifer She, Parsa Mahmoudieh, Aliya Rysbek, Wei-Jen Ko, Andrea Huber, Brett Wiltshire, Gal Elidan, Roni Rabin, Jasmin Rubinovitz, Amit Pitaru, Mac McAllister, Julia Wilkowski, David Choi, Roee Engelberg, Lidan Hackmon, Adva Levin, Rachel Griffin, Michael Sears, Filip Bar, Mia Mesar, Mana Jabbour, Arslan Chaudhry, James Cohan, Sridhar Thiagarajan, Nir Levine, Ben Brown, Dilan Gorur, Svetlana Grant, Rachel Hashimshoni, Laura Weidinger, Jieru Hu, Dawn Chen, Kuba Dolecki, Canfer Akbulut, Maxwell Bileschi, Laura Culp, Wen-Xin Dong, Nahema Marchal, Kelsie Van Deman, Hema Bajaj Misra, Michael Duah, Moran Ambar, Avi Caciularu, Sandra Lefdal, Chris Summerfield, James An, Pierre-Alexandre Kamienny, Abhinit Mohdi, Theofilos Strinopoulous, Annie Hale, Wayne Anderson, Luis C. Cobo, Niv Efron, Muktha Ananda, Shakir Mohamed, Maureen Heymans, Zoubin Ghahramani, Yossi Matias, Ben Gomes, Lila Ibrahim
Abstract: A major challenge facing the world is the provision of equitable and universal access to quality education. Recent advances in generative AI (gen AI) have created excitement about the potential of new technologies to offer a personal tutor for every learner and a teaching assistant for every teacher. The full extent of this dream, however, has not yet materialised. We argue that this is primarily due to the difficulties with verbalising pedagogical intuitions into gen AI prompts and the lack of good evaluation practices, reinforced by the challenges in defining excellent pedagogy. Here we present our work collaborating with learners and educators to translate high level principles from learning science into a pragmatic set of seven diverse educational benchmarks, spanning quantitative, qualitative, automatic and human evaluations; and to develop a new set of fine-tuning datasets to improve the pedagogical capabilities of Gemini, introducing LearnLM-Tutor. Our evaluations show that LearnLM-Tutor is consistently preferred over a prompt tuned Gemini by educators and learners on a number of pedagogical dimensions. We hope that this work can serve as a first step towards developing a comprehensive educational evaluation framework, and that this can enable rapid progress within the AI and EdTech communities towards maximising the positive impact of gen AI in education.
Authors: Irina Abdullaeva, Andrei Filatov, Mikhail Orlov, Ivan Karpukhin, Viacheslav Vasilev, Denis Dimitrov, Andrey Kuznetsov, Ivan Kireev, Andrey Savchenko
Abstract: Event sequences (ESs) arise in many practical domains including finance, retail, social networks, and healthcare. In the context of machine learning, event sequences can be seen as a special type of tabular data with annotated timestamps. Despite the importance of ESs modeling and analysis, little effort was made in adapting large language models (LLMs) to the ESs domain. In this paper, we highlight the common difficulties of ESs processing and propose a novel solution capable of solving multiple downstream tasks with little or no finetuning. In particular, we solve the problem of working with long sequences and improve time and numeric features processing. The resulting method, called ESQA, effectively utilizes the power of LLMs and, according to extensive experiments, achieves state-of-the-art results in the ESs domain.
Authors: Jianan Fan, Dongnan Liu, Canran Li, Hang Chang, Heng Huang, Filip Braet, Mei Chen, Weidong Cai
Abstract: Cellular nuclei recognition serves as a fundamental and essential step in the workflow of digital pathology. However, with disparate source organs and staining procedures among histology image clusters, the scanned tiles inherently conform to a non-uniform data distribution, which induces deteriorated promises for general cross-cohort usages. Despite the latest efforts leveraging domain adaptation to mitigate distributional discrepancy, those methods are subjected to modeling the morphological characteristics of each cell individually, disregarding the hierarchical latent structure and intrinsic contextual correspondences across the tumor micro-environment. In this work, we identify the importance of implicit correspondences across biological contexts for exploiting domain-invariant pathological composition and thereby propose to exploit the dependence over various biological structures for domain adaptive cellular recognition. We discover those high-level correspondences via unsupervised contextual modeling and use them as bridges to facilitate adaptation over diverse organs and stains. In addition, to further exploit the rich spatial contexts embedded amongst nuclear communities, we propose self-adaptive dynamic distillation to secure instance-aware trade-offs across different model constituents. The proposed method is extensively evaluated on a broad spectrum of cross-domain settings under miscellaneous data distribution shifts and outperforms the state-of-the-art methods by a substantial margin. Code is available at https://github.com/camwew/CellularRecognition_DA_CC.
Authors: Linping Qu, Yuyi Mao, Shenghui Song, Chi-Ying Tsui
Abstract: One of the most critical challenges for deploying distributed learning solutions, such as federated learning (FL), in wireless networks is the limited battery capacity of mobile clients. While it is a common belief that the major energy consumption of mobile clients comes from the uplink data transmission, this paper presents a novel finding, namely the channel decoding operation also contributes significantly to the overall energy consumption of mobile clients in FL. Motivated by this new observation, we propose an energy-efficient adaptive channel decoding scheme that leverages the intrinsic robustness of FL to model errors. In particular, the robustness is exploited to reduce the energy consumption of channel decoders at mobile clients by adaptively adjusting the number of decoding iterations. We theoretically prove that wireless FL with communication errors can converge at the same rate as the case with error-free communication as long as the bit error rate (BER) is properly constrained. An adaptive channel decoding scheme is then proposed to improve the energy efficiency of wireless FL systems. Experimental results demonstrate that the proposed method maintains the same learning accuracy while reducing the channel decoding energy consumption by 20% when compared to existing approaches.