Authors: Yaguang Song, Xiaoshan Yang, Dongmei Jiang, Yaowei Wang, Changsheng Xu
Abstract: Incremental learning aims to enable models to continuously acquire knowledge from evolving data streams while preserving previously learned capabilities. While current research predominantly focuses on unimodal incremental learning and multimodal incremental learning where the modalities are consistent, real-world scenarios often present data from entirely new modalities, posing additional challenges. This paper investigates the feasibility of developing a unified model capable of incremental learning across continuously evolving modal sequences. To this end, we introduce a novel paradigm called Modality Incremental Learning (MIL), where each learning stage involves data from distinct modalities. To address this task, we propose a novel framework named Harmony, designed to achieve modal alignment and knowledge retention, enabling the model to reduce the modal discrepancy and learn from a sequence of distinct modalities, ultimately completing tasks across multiple modalities within a unified framework. Our approach introduces the adaptive compatible feature modulation and cumulative modal bridging. Through constructing historical modal features and performing modal knowledge accumulation and alignment, the proposed components collaboratively bridge modal differences and maintain knowledge retention, even with solely unimodal data available at each learning phase.These components work in concert to establish effective modality connections and maintain knowledge retention, even when only unimodal data is available at each learning stage. Extensive experiments on the MIL task demonstrate that our proposed method significantly outperforms existing incremental learning methods, validating its effectiveness in MIL scenarios.
Authors: Wenxuan Yang, Qingqu Wei, Chenxi Ma, Weimin Tan, Bo Yan
Abstract: Current scaling laws for visual AI models focus predominantly on large-scale pretraining, leaving a critical gap in understanding how performance scales for data-constrained downstream tasks. To address this limitation, this paper establishes the first practical framework for data-efficient scaling laws in visual transfer learning, addressing two fundamental questions: 1) How do scaling behaviors shift when downstream tasks operate with limited data? 2) What governs the efficacy of knowledge distillation under such constraints? Through systematic analysis of vision tasks across data regimes (1K-1M samples), we propose the distillation boundary theory, revealing a critical turning point in distillation efficiency: 1) Distillation superiority: In data-scarce conditions, distilled models significantly outperform their non-distillation counterparts, efficiently leveraging inherited knowledge to compensate for limited training samples. 2) Pre-training dominance: As pre-training data increases beyond a critical threshold, non-distilled models gradually surpass distilled versions, suggesting diminishing returns from knowledge inheritance when sufficient task-specific data becomes available. Empirical validation across various model scales (2.5M to 38M parameters) and data volumes demonstrate these performance inflection points, with error difference curves transitioning from positive to negative values at critical data thresholds, confirming our theoretical predictions. This work redefines scaling laws for data-limited regimes, bridging the knowledge gap between large-scale pretraining and practical downstream adaptation, addressing a critical barrier to understanding vision model scaling behaviors and optimizing computational resource allocation.
Authors: Anna C. M. Th\"oni, Yoram Bachrach, Tal Kachman
Abstract: Mean-field game theory relies on approximating games that would otherwise have been intractable to model. While the games can be solved analytically via the associated system of partial derivatives, this approach is not model-free, can lead to the loss of the existence or uniqueness of solutions and may suffer from modelling bias. To reduce the dependency between the model and the game, we combine mean-field game theory with deep learning in the form of neural ordinary differential equations. The resulting model is data-driven, lightweight and can learn extensive strategic interactions that are hard to capture using mean-field theory alone. In addition, the model is based on automatic differentiation, making it more robust and objective than approaches based on finite differences. We highlight the efficiency and flexibility of our approach by solving three mean-field games that vary in their complexity, observability and the presence of noise. Using these results, we show that the model is flexible, lightweight and requires few observations to learn the distribution underlying the data.
Authors: Yifei Wang, Qi Liu, Fuli Min, Honghao Wang
Abstract: Polysomnography (PSG) signals are essential for studying sleep processes and diagnosing sleep disorders. Analyzing PSG data through deep neural networks (DNNs) for automated sleep monitoring has become increasingly feasible. However, the limited availability of datasets for certain sleep events often leads to DNNs focusing on a single task with a single-sourced training dataset. As a result, these models struggle to transfer to new sleep events and lack robustness when applied to new datasets. To address these challenges, we propose PSG-MAE, a mask autoencoder (MAE) based pre-training framework. By performing self-supervised learning on a large volume of unlabeled PSG data, PSG-MAE develops a robust feature extraction network that can be broadly applied to various sleep event monitoring tasks. Unlike conventional MAEs, PSG-MAE generates complementary masks across PSG channels, integrates a multichannel signal reconstruction method, and employs a self-supervised inter-channel contrastive learning (ICCL) strategy. This approach enables the encoder to capture temporal features from each channel while simultaneously learning latent relationships between channels, thereby enhancing the utilization of multichannel information. Experimental results show that PSG-MAE effectively captures both temporal details and inter-channel information from PSG signals. When the encoder pre-trained through PSG-MAE is fine-tuned with downstream feature decomposition networks, it achieves an accuracy of 83.7% for sleep staging and 90.45% for detecting obstructive sleep apnea, which highlights the framework's robustness and broad applicability.
Authors: Alireza Rafiei, Gari D. Clifford, Nasim Katebi
Abstract: Fetal health monitoring through one-dimensional Doppler ultrasound (DUS) signals offers a cost-effective and accessible approach that is increasingly gaining interest. Despite its potential, the development of machine learning based techniques to assess the health condition of mothers and fetuses using DUS signals remains limited. This scarcity is primarily due to the lack of extensive DUS datasets with a reliable reference for interpretation and data imbalance across different gestational ages. In response, we introduce a novel autoregressive generative model designed to map fetal electrocardiogram (FECG) signals to corresponding DUS waveforms (Auto-FEDUS). By leveraging a neural temporal network based on dilated causal convolutions that operate directly on the waveform level, the model effectively captures both short and long-range dependencies within the signals, preserving the integrity of generated data. Cross-subject experiments demonstrate that Auto-FEDUS outperforms conventional generative architectures across both time and frequency domain evaluations, producing DUS signals that closely resemble the morphology of their real counterparts. The realism of these synthesized signals was further gauged using a quality assessment model, which classified all as good quality, and a heart rate estimation model, which produced comparable results for generated and real data, with a Bland-Altman limit of 4.5 beats per minute. This advancement offers a promising solution for mitigating limited data availability and enhancing the training of DUS-based fetal models, making them more effective and generalizable.
Authors: Hanyu Zhang, Zhen Xing, Wenxuan Yang, Chenxi Ma, Weimin Tan, Bo Yan
Abstract: As transfer learning models and datasets grow larger, efficient adaptation and storage optimization have become critical needs. Coreset selection addresses these challenges by identifying and retaining the most informative samples, constructing a compact subset for target domain training. However, current methods primarily rely on instance-level difficulty assessments, overlooking crucial category-level characteristics and consequently under-representing minority classes. To overcome this limitation, we propose Non-Uniform Class-Wise Coreset Selection (NUCS), a novel framework that integrates both class-level and instance-level criteria. NUCS automatically allocates data selection budgets for each class based on intrinsic category difficulty and adaptively selects samples within optimal difficulty ranges. By explicitly incorporating category-specific insights, our approach achieves a more balanced and representative coreset, addressing key shortcomings of prior methods. Comprehensive theoretical analysis validates the rationale behind adaptive budget allocation and sample selection, while extensive experiments across 14 diverse datasets and model architectures demonstrate NUCS's consistent improvements over state-of-the-art methods, achieving superior accuracy and computational efficiency. Notably, on CIFAR100 and Food101, NUCS matches full-data training accuracy while retaining just 30% of samples and reducing computation time by 60%. Our work highlights the importance of characterizing category difficulty in coreset selection, offering a robust and data-efficient solution for transfer learning.
Authors: Aleksandr Mikhalev, Aleksandr Katrutsa, Konstantin Sozykin, Ivan Oseledets
Abstract: This study presents an NNTile framework for training large deep neural networks in heterogeneous clusters. The NNTile is based on a StarPU library, which implements task-based parallelism and schedules all provided tasks onto all available processing units (CPUs and GPUs). It means that a particular operation, necessary to train a large neural network, can be performed on any of the CPU cores or GPU devices, depending on automatic scheduling decisions. Such an approach shifts the burden of deciding where to compute and when to communicate from a human being to an automatic decision maker, whether a simple greedy heuristic or a complex AI-based software. The performance of the presented tool for training large language models is demonstrated in extensive numerical experiments.
Authors: Paul Ghanem, Michael Potter, Owen Howell, Pau Closas, Alireza Ramezani, Deniz Erdogmus, Robert Platt, Tales Imbiriba
Abstract: Inferring an adversary's goals from exhibited behavior is crucial for counterplanning and non-cooperative multi-agent systems in domains like cybersecurity, military, and strategy games. Deep Inverse Reinforcement Learning (IRL) methods based on maximum entropy principles show promise in recovering adversaries' goals but are typically offline, require large batch sizes with gradient descent, and rely on first-order updates, limiting their applicability in real-time scenarios. We propose an online Recursive Deep Inverse Reinforcement Learning (RDIRL) approach to recover the cost function governing the adversary actions and goals. Specifically, we minimize an upper bound on the standard Guided Cost Learning (GCL) objective using sequential second-order Newton updates, akin to the Extended Kalman Filter (EKF), leading to a fast (in terms of convergence) learning algorithm. We demonstrate that RDIRL is able to recover cost and reward functions of expert agents in standard and adversarial benchmark tasks. Experiments on benchmark tasks show that our proposed approach outperforms several leading IRL algorithms.
Authors: Zichao Yue, Chenhui Deng, Zhiru Zhang
Abstract: Graph neural networks (GNNs) are widely used for learning node embeddings in graphs, typically adopting a message-passing scheme. This approach, however, leads to the neighbor explosion problem, with exponentially growing computational and memory demands as layers increase. Graph sampling has become the predominant method for scaling GNNs to large graphs, mitigating but not fully solving the issue. Pre-propagation GNNs (PP-GNNs) represent a new class of models that decouple feature propagation from training through pre-processing, addressing neighbor explosion in theory. Yet, their practical advantages and system-level optimizations remain underexplored. This paper provides a comprehensive characterization of PP-GNNs, comparing them with graph-sampling-based methods in training efficiency, scalability, and accuracy. While PP-GNNs achieve comparable accuracy, we identify data loading as the key bottleneck for training efficiency and input expansion as a major scalability challenge. To address these issues, we propose optimized data loading schemes and tailored training methods that improve PP-GNN training throughput by an average of 15$\times$ over the PP-GNN baselines, with speedup of up to 2 orders of magnitude compared to sampling-based GNNs on large graph benchmarks. Our implementation is publicly available at https://github.com/cornell-zhang/preprop-gnn.
Authors: Zhiwei Xu, Zhiyu Ni, Yixin Wang, Wei Hu
Abstract: ''Grokking'' is a phenomenon where a neural network first memorizes training data and generalizes poorly, but then suddenly transitions to near-perfect generalization after prolonged training. While intriguing, this delayed generalization phenomenon compromises predictability and efficiency. Ideally, models should generalize directly without delay. To this end, this paper proposes GrokTransfer, a simple and principled method for accelerating grokking in training neural networks, based on the key observation that data embedding plays a crucial role in determining whether generalization is delayed. GrokTransfer first trains a smaller, weaker model to reach a nontrivial (but far from optimal) test performance. Then, the learned input embedding from this weaker model is extracted and used to initialize the embedding in the target, stronger model. We rigorously prove that, on a synthetic XOR task where delayed generalization always occurs in normal training, GrokTransfer enables the target model to generalize directly without delay. Moreover, we demonstrate that, across empirical studies of different tasks, GrokTransfer effectively reshapes the training dynamics and eliminates delayed generalization, for both fully-connected neural networks and Transformers.
Authors: Ganesh Sundaram, Jonas Ulmen, Daniel G\"orges
Abstract: Deep neural networks (DNNs) deliver outstanding performance, but their complexity often prohibits deployment in resource-constrained settings. Comprehensive structured pruning frameworks based on parameter dependency analysis reduce model size with specific regard to computational performance. When applying them to Multi-Component Neural Architectures (MCNAs), they risk network integrity by removing large parameter groups. We introduce a component-aware pruning strategy, extending dependency graphs to isolate individual components and inter-component flows. This creates smaller, targeted pruning groups that conserve functional integrity. Demonstrated effectively on a control task, our approach achieves greater sparsity and reduced performance degradation, opening a path for optimizing complex, multi-component DNNs efficiently.
Authors: Ibrahim Emirahmetoglu, David E. Stewart
Abstract: Hessian-free (HF) optimization has been shown to effectively train deep autoencoders (Martens, 2010). In this paper, we aim to accelerate HF training of autoencoders by reducing the amount of data used in training. HF utilizes the conjugate gradient algorithm to estimate update directions. Instead, we propose using the LSMR method, which is known for effectively solving large sparse linear systems. We also incorporate Chapelle & Erhan (2011)'s improved preconditioner for HF optimization. In addition, we introduce a new mini-batch selection algorithm to mitigate overfitting. Our algorithm starts with a small subset of the training data and gradually increases the mini-batch size based on (i) variance estimates obtained during the computation of a mini-batch gradient (Byrd et al., 2012) and (ii) the relative decrease in objective value for the validation data. Our experimental results demonstrate that our stochastic Hessian-free optimization, using the LSMR method and the new sample selection algorithm, leads to rapid training of deep autoencoders with improved generalization error.
Authors: Yassine Ouzar, Cl\'emence Nineuil, Fouad Boutaleb, Emery Pierson, Ali Amad, Mohamed Daoudi
Abstract: Depression is a complex mental disorder characterized by a diverse range of observable and measurable indicators that go beyond traditional subjective assessments. Recent research has increasingly focused on objective, passive, and continuous monitoring using wearable devices to gain more precise insights into the physiological and behavioral aspects of depression. However, most existing studies primarily distinguish between healthy and depressed individuals, adopting a binary classification that fails to capture the heterogeneity of depressive disorders. In this study, we leverage wearable devices to predict depression subtypes-specifically unipolar and bipolar depression-aiming to identify distinctive biomarkers that could enhance diagnostic precision and support personalized treatment strategies. To this end, we introduce the CALYPSO dataset, designed for non-invasive detection of depression subtypes and symptomatology through physiological and behavioral signals, including blood volume pulse, electrodermal activity, body temperature, and three-axis acceleration. Additionally, we establish a benchmark on the dataset using well-known features and standard machine learning methods. Preliminary results indicate that features related to physical activity, extracted from accelerometer data, are the most effective in distinguishing between unipolar and bipolar depression, achieving an accuracy of $96.77\%$. Temperature-based features also showed high discriminative power, reaching an accuracy of $93.55\%$. These findings highlight the potential of physiological and behavioral monitoring for improving the classification of depressive subtypes, paving the way for more tailored clinical interventions.
Authors: Omid Sedehi, Manish Yadav, Merten Stender, Sebastian Oberst
Abstract: Measurements acquired from distributed physical systems are often sparse and noisy. Therefore, signal processing and system identification tools are required to mitigate noise effects and reconstruct unobserved dynamics from limited sensor data. However, this process is particularly challenging because the fundamental equations governing the dynamics are largely unavailable in practice. Reservoir Computing (RC) techniques have shown promise in efficiently simulating dynamical systems through an unstructured and efficient computation graph comprising a set of neurons with random connectivity. However, the potential of RC to operate in noisy regimes and distinguish noise from the primary dynamics of the system has not been fully explored. This paper presents a novel RC method for noise filtering and reconstructing nonlinear dynamics, offering a novel learning protocol associated with hyperparameter optimization. The performance of the RC in terms of noise intensity, noise frequency content, and drastic shifts in dynamical parameters are studied in two illustrative examples involving the nonlinear dynamics of the Lorenz attractor and adaptive exponential integrate-and-fire system (AdEx). It is shown that the denoising performance improves via truncating redundant nodes and edges of the computing reservoir, as well as properly optimizing the hyperparameters, e.g., the leakage rate, the spectral radius, the input connectivity, and the ridge regression parameter. Furthermore, the presented framework shows good generalization behavior when tested for reconstructing unseen attractors from the bifurcation diagram. Compared to the Extended Kalman Filter (EKF), the presented RC framework yields competitive accuracy at low signal-to-noise ratios (SNRs) and high-frequency ranges.
Authors: Haoran Xu, Shuozhe Li, Harshit Sikchi, Scott Niekum, Amy Zhang
Abstract: We introduce Iterative Dual Reinforcement Learning (IDRL), a new method that takes an optimal discriminator-weighted imitation view of solving RL. Our method is motivated by a simple experiment in which we find training a discriminator using the offline dataset plus an additional expert dataset and then performing discriminator-weighted behavior cloning gives strong results on various types of datasets. That optimal discriminator weight is quite similar to the learned visitation distribution ratio in Dual-RL, however, we find that current Dual-RL methods do not correctly estimate that ratio. In IDRL, we propose a correction method to iteratively approach the optimal visitation distribution ratio in the offline dataset given no addtional expert dataset. During each iteration, IDRL removes zero-weight suboptimal transitions using the learned ratio from the previous iteration and runs Dual-RL on the remaining subdataset. This can be seen as replacing the behavior visitation distribution with the optimized visitation distribution from the previous iteration, which theoretically gives a curriculum of improved visitation distribution ratios that are closer to the optimal discriminator weight. We verify the effectiveness of IDRL on various kinds of offline datasets, including D4RL datasets and more realistic corrupted demonstrations. IDRL beats strong Primal-RL and Dual-RL baselines in terms of both performance and stability, on all datasets.
Authors: Yegor Klochkov
Abstract: One of the goals of language model unlearning is to reduce memorization of selected text instances while retaining the model's general abilities. Despite various proposed methods, reducing memorization of large datasets without noticeable degradation in model utility remains challenging. In this paper, we investigate the mean teacher algorithm (Tarvainen & Valpola, 2017), a simple proximal optimization method from continual learning literature that gradually modifies the teacher model. We show that the mean teacher can approximate a trajectory of a slow natural gradient descent (NGD), which inherently seeks low-curvature updates that are less likely to degrade the model utility. While slow NGD can suffer from vanishing gradients, we introduce a new unlearning loss called "negative log-unlikelihood" (NLUL) that avoids this problem. We show that the combination of mean teacher and NLUL improves some metrics on the MUSE benchmarks (Shi et al., 2024).
Authors: Haldun Balim, Yang Hu, Yuyang Zhang, Na Li
Abstract: Imitation learning is a widely used approach for training agents to replicate expert behavior in complex decision-making tasks. However, existing methods often struggle with compounding errors and limited generalization, due to the inherent challenge of error correction and the distribution shift between training and deployment. In this paper, we present a novel model-based imitation learning framework inspired by model predictive control, which addresses these limitations by integrating predictive modeling through multi-step state predictions. Our method outperforms traditional behavior cloning numerical benchmarks, demonstrating superior robustness to distribution shift and measurement noise both in available data and during execution. Furthermore, we provide theoretical guarantees on the sample complexity and error bounds of our method, offering insights into its convergence properties.
Authors: Saksham Rastogi, Pratyush Maini, Danish Pruthi
Abstract: Given how large parts of publicly available text are crawled to pretrain large language models (LLMs), data creators increasingly worry about the inclusion of their proprietary data for model training without attribution or licensing. Their concerns are also shared by benchmark curators whose test-sets might be compromised. In this paper, we present STAMP, a framework for detecting dataset membership-i.e., determining the inclusion of a dataset in the pretraining corpora of LLMs. Given an original piece of content, our proposal involves first generating multiple rephrases, each embedding a watermark with a unique secret key. One version is to be released publicly, while others are to be kept private. Subsequently, creators can compare model likelihoods between public and private versions using paired statistical tests to prove membership. We show that our framework can successfully detect contamination across four benchmarks which appear only once in the training data and constitute less than 0.001% of the total tokens, outperforming several contamination detection and dataset inference baselines. We verify that STAMP preserves both the semantic meaning and the utility of the original data in comparing different models. We apply STAMP to two real-world scenarios to confirm the inclusion of paper abstracts and blog articles in the pretraining corpora.
Authors: Vivek Oommen, Andreas E. Robertson, Daniel Diaz, Coleman Alleman, Zhen Zhang, Anthony D. Rollett, George E. Karniadakis, R\'emi Dingreville
Abstract: Neural surrogate solvers can estimate solutions to partial differential equations in physical problems more efficiently than standard numerical methods, but require extensive high-resolution training data. In this paper, we break this limitation; we introduce a framework for super-resolution learning in solid mechanics problems. Our approach allows one to train a high-resolution neural network using only low-resolution data. Our Equilibrium Conserving Operator (ECO) architecture embeds known physics directly into the network to make up for missing high-resolution information during training. We evaluate this ECO-based super-resolution framework that strongly enforces conservation-laws in the predicted solutions on two working examples: embedded pores in a homogenized matrix and randomly textured polycrystalline materials. ECO eliminates the reliance on high-fidelity data and reduces the upfront cost of data collection by two orders of magnitude, offering a robust pathway for resource-efficient surrogate modeling in materials modeling. ECO is readily generalizable to other physics-based problems.
Authors: Jielong LuZhihao Wu, Zhiling Cai, Yueyang Pi, Shiping Wang
Abstract: In recent years, Graph Convolutional Networks (GCNs) have gained popularity for their exceptional ability to process graph-structured data. Existing GCN-based approaches typically employ a shallow model architecture due to the over-smoothing phenomenon. Current approaches to mitigating over-smoothing primarily involve adding supplementary components to GCN architectures, such as residual connections and random edge-dropping strategies. However, these improvements toward deep GCNs have achieved only limited success. In this work, we analyze the intrinsic message passing mechanism of GCNs and identify a critical issue: messages originating from high-order neighbors must traverse through low-order neighbors to reach the target node. This repeated reliance on low-order neighbors leads to redundant information aggregation, a phenomenon we term over-aggregation. Our analysis demonstrates that over-aggregation not only introduces significant redundancy but also serves as the fundamental cause of over-smoothing in GCNs.
Authors: Shenzhi Yang, Bin Liang, An Liu, Lin Gui, Xingkai Yao, Xiaofang Zhang
Abstract: Given the critical role of graphs in real-world applications and their high-security requirements, improving the ability of graph neural networks (GNNs) to detect out-of-distribution (OOD) data is an urgent research problem. The recent work GNNSAFE proposes a framework based on the aggregation of negative energy scores that significantly improves the performance of GNNs to detect node-level OOD data. However, our study finds that score aggregation among nodes is susceptible to extreme values due to the unboundedness of the negative energy scores and logit shifts, which severely limits the accuracy of GNNs in detecting node-level OOD data. In this paper, we propose NODESAFE: reducing the generation of extreme scores of nodes by adding two optimization terms that make the negative energy scores bounded and mitigate the logit shift. Experimental results show that our approach dramatically improves the ability of GNNs to detect OOD data at the node level, e.g., in detecting OOD data induced by Structure Manipulation, the metric of FPR95 (lower is better) in scenarios without (with) OOD data exposure are reduced from the current SOTA by 28.4% (22.7%).
Authors: Vasista Ramachandruni, Sai Hruday Reddy Nara, Geo Lalu, Sabrina Yang, Mohit Ramesh Kumar, Aarjav Jain, Pratham Mehta, Hankyu Koo, Jason Damonte, Marx Akl
Abstract: A chaotic system is a highly volatile system characterized by its sensitive dependence on initial conditions and outside factors. Chaotic systems are prevalent throughout the world today: in weather patterns, disease outbreaks, and even financial markets. Chaotic systems are seen in every field of science and humanities, so being able to predict these systems is greatly beneficial to society. In this study, we evaluate 10 different machine learning models and neural networks [1] based on Root Mean Squared Error (RMSE) and R^2 values for their ability to predict one of these systems, the multi-pendulum. We begin by generating synthetic data representing the angles of the pendulum over time using the Runge Kutta Method for solving 4th Order Differential Equations (ODE-RK4) [2]. At first, we used the single-step sliding window approach, predicting the 50st step after training for steps 0-49 and so forth. However, to more accurately cover chaotic motion and behavior in these systems, we transitioned to a time-step based approach. Here, we trained the model/network on many initial angles and tested it on a completely new set of initial angles, or 'in-between' to capture chaotic motion to its fullest extent. We also evaluated the stability of the system using Lyapunov exponents. We concluded that for a double pendulum, the best model was the Long Short Term Memory Network (LSTM)[3] for the sliding window and time step approaches in both friction and frictionless scenarios. For triple pendulum, the Vanilla Recurrent Neural Network (VRNN)[4] was the best for the sliding window and Gated Recurrent Network (GRU) [5] was the best for the time step approach, but for friction, LSTM was the best.
Authors: Hui Yeok Wong, Chee Kau Lim, Chee Seng Chan
Abstract: Federated Learning (FL) on non-independently and identically distributed (non-IID) data remains a critical challenge, as existing approaches struggle with severe data heterogeneity. Current methods primarily address symptoms of non-IID by applying incremental adjustments to Federated Averaging (FedAvg), rather than directly resolving its inherent design limitations. Consequently, performance significantly deteriorates under highly heterogeneous conditions, as the fundamental issue of imbalanced exposure to diverse class and feature distributions remains unresolved. This paper introduces Stratify, a novel FL framework designed to systematically manage class and feature distributions throughout training, effectively tackling the root cause of non-IID challenges. Inspired by classical stratified sampling, our approach employs a Stratified Label Schedule (SLS) to ensure balanced exposure across labels, significantly reducing bias and variance in aggregated gradients. Complementing SLS, we propose a label-aware client selection strategy, restricting participation exclusively to clients possessing data relevant to scheduled labels. Additionally, Stratify incorporates a fine-grained, high-frequency update scheme, accelerating convergence and further mitigating data heterogeneity. To uphold privacy, we implement a secure client selection protocol leveraging homomorphic encryption, enabling precise global label statistics without disclosing sensitive client information. Extensive evaluations on MNIST, CIFAR-10, CIFAR-100, Tiny-ImageNet, COVTYPE, PACS, and Digits-DG demonstrate that Stratify attains performance comparable to IID baselines, accelerates convergence, and reduces client-side computation compared to state-of-the-art methods, underscoring its practical effectiveness in realistic federated learning scenarios.
Authors: Duy A. Nguyen, Quan Huu Do, Khoa D. Doan, Minh N. Do
Abstract: Multimodal learning has demonstrated incredible successes by integrating diverse data sources, yet it often relies on the availability of all modalities - an assumption that rarely holds in real-world applications. Pretrained multimodal models, while effective, struggle when confronted with small-scale and incomplete datasets (i.e., missing modalities), limiting their practical applicability. Previous studies on reconstructing missing modalities have overlooked the reconstruction's potential unreliability, which could compromise the quality of the final outputs. We present SURE (Scalable Uncertainty and Reconstruction Estimation), a novel framework that extends the capabilities of pretrained multimodal models by introducing latent space reconstruction and uncertainty estimation for both reconstructed modalities and downstream tasks. Our method is architecture-agnostic, reconstructs missing modalities, and delivers reliable uncertainty estimates, improving both interpretability and performance. SURE introduces a unique Pearson Correlation-based loss and applies statistical error propagation in deep networks for the first time, allowing precise quantification of uncertainties from missing data and model predictions. Extensive experiments across tasks such as sentiment analysis, genre classification, and action recognition show that SURE consistently achieves state-of-the-art performance, ensuring robust predictions even in the presence of incomplete data.
Authors: Jiadong Lou, Bingqing Liu, Yuanheng Xiong, Xiaodong Zhang, Xu Yuan
Abstract: Phytoplankton absorb and scatter light in unique ways, subtly altering the color of water, changes that are often minor for human eyes to detect but can be captured by sensitive ocean color instruments onboard satellites from space. Hyperspectral sensors, paired with advanced algorithms, are expected to significantly enhance the characterization of phytoplankton community composition, especially in coastal waters where ocean color remote sensing applications have historically encountered significant challenges. This study presents novel machine learning-based solutions for NASA's hyperspectral missions, including EMIT and PACE, tackling high-fidelity retrievals of phytoplankton absorption coefficient and chlorophyll a from their hyperspectral remote sensing reflectance. Given that a single Rrs spectrum may correspond to varied combinations of inherent optical properties and associated concentrations, the Variational Autoencoder (VAE) is used as a backbone in this study to handle such multi-distribution prediction problems. We first time tailor the VAE model with innovative designs to achieve hyperspectral retrievals of aphy and of Chl-a from hyperspectral Rrs in optically complex estuarine-coastal waters. Validation with extensive experimental observation demonstrates superior performance of the VAE models with high precision and low bias. The in-depth analysis of VAE's advanced model structures and learning designs highlights the improvement and advantages of VAE-based solutions over the mixture density network (MDN) approach, particularly on high-dimensional data, such as PACE. Our study provides strong evidence that current EMIT and PACE hyperspectral data as well as the upcoming Surface Biology Geology mission will open new pathways toward a better understanding of phytoplankton community dynamics in aquatic ecosystems when integrated with AI technologies.
Authors: Vivian Lin, Ramneet Kaur, Yahan Yang, Souradeep Dutta, Yiannis Kantaros, Anirban Roy, Susmit Jha, Oleg Sokolsky, Insup Lee
Abstract: The safety of learning-enabled cyber-physical systems is compromised by the well-known vulnerabilities of deep neural networks to out-of-distribution (OOD) inputs. Existing literature has sought to monitor the safety of such systems by detecting OOD data. However, such approaches have limited utility, as the presence of an OOD input does not necessarily imply the violation of a desired safety property. We instead propose to directly monitor safety in a manner that is itself robust to OOD data. To this end, we predict violations of signal temporal logic safety specifications based on predicted future trajectories. Our safety monitor additionally uses a novel combination of adaptive conformal prediction and incremental learning. The former obtains probabilistic prediction guarantees even on OOD data, and the latter prevents overly conservative predictions. We evaluate the efficacy of the proposed approach in two case studies on safety monitoring: 1) predicting collisions of an F1Tenth car with static obstacles, and 2) predicting collisions of a race car with multiple dynamic obstacles. We find that adaptive conformal prediction obtains theoretical guarantees where other uncertainty quantification methods fail to do so. Additionally, combining adaptive conformal prediction and incremental learning for safety monitoring achieves high recall and timeliness while reducing loss in precision. We achieve these results even in OOD settings and outperform alternative methods.
Authors: Minsu Koh, Beom-Chul Park, Heejo Kong, Seong-Whan Lee
Abstract: Neural operators have emerged as promising frameworks for learning mappings governed by partial differential equations (PDEs), serving as data-driven alternatives to traditional numerical methods. While methods such as the Fourier neural operator (FNO) have demonstrated notable performance, their reliance on uniform grids restricts their applicability to complex geometries and irregular meshes. Recently, Transformer-based neural operators with linear attention mechanisms have shown potential in overcoming these limitations for large-scale PDE simulations. However, these approaches predominantly emphasize global feature aggregation, often overlooking fine-scale dynamics and localized PDE behaviors essential for accurate solutions. To address these challenges, we propose the Locality-Aware Attention Transformer (LA2Former), which leverages K-nearest neighbors for dynamic patchifying and integrates global-local attention for enhanced PDE modeling. By combining linear attention for efficient global context encoding with pairwise attention for capturing intricate local interactions, LA2Former achieves an optimal balance between computational efficiency and predictive accuracy. Extensive evaluations across six benchmark datasets demonstrate that LA2Former improves predictive accuracy by over 50% relative to existing linear attention methods, while also outperforming full pairwise attention under optimal conditions. This work underscores the critical importance of localized feature learning in advancing Transformer-based neural operators for solving PDEs on complex and irregular domains.
Authors: Yiran Wang, Tangtang Xie, Hao Wu
Abstract: Non-Intrusive Load Monitoring (NILM) has emerged as a key smart grid technology, identifying electrical device and providing detailed energy consumption data for precise demand response management. Nevertheless, NILM data suffers from missing values due to inescapable factors like sensor failure, leading to inaccuracies in non-intrusive load monitoring. A stochastic gradient descent (SGD)-based latent factorization of tensors model has proven to be effective in estimating missing data, however, it updates a latent factor solely based on the current stochastic gradient, without considering past information, which leads to slow convergence of anLFT model. To address this issue, this paper proposes a Nonlinear Proportional-integral-derivative (PID)-Incorporated Latent factorization of tensors (NPIL) model with two-fold ideas: a) rebuilding the instant learning error according to the principle of a nonlinear PID controller, thus, the past update information is efficiently incorporated into the learning scheme, and b) implementing gain parameter adaptation by utilizing particle swarm optimization (PSO) algorithm, hence, the model computational efficiency is effectively improved. Experimental results on real-world NILM datasets demonstrate that the proposed NPIL model surpasses state-of-the-art models in convergence rate and accuracy when predicting the missing NILM data.
Authors: Vivian Lin, Insup Lee
Abstract: With the known vulnerability of neural networks to distribution shift, maintaining reliability in learning-enabled cyber-physical systems poses a salient challenge. In response, many existing methods adopt a detect and abstain methodology, aiming to detect distribution shift at inference time so that the learning-enabled component can abstain from decision-making. This approach, however, has limited use in real-world applications. We instead propose a monitor and recover paradigm as a promising direction for future research. This philosophy emphasizes 1) robust safety monitoring instead of distribution shift detection and 2) distribution shift recovery instead of abstention. We discuss two examples from our recent work.
Authors: Kasymkhan Khubiev, Michail Semenov
Abstract: Algorithmic trading relies on extracting meaningful signals from diverse financial data sources, including candlestick charts, order statistics on put and canceled orders, traded volume data, limit order books, and news flow. While deep learning has demonstrated remarkable success in processing unstructured data and has significantly advanced natural language processing, its application to structured financial data remains an ongoing challenge. This study investigates the integration of deep learning models with financial data modalities, aiming to enhance predictive performance in trading strategies and portfolio optimization. We present a novel approach to incorporating limit order book analysis into algorithmic trading by developing embedding techniques and treating sequential limit order book snapshots as distinct input channels in an image-based representation. Our methodology for processing limit order book data achieves state-of-the-art performance in high-frequency trading algorithms, underscoring the effectiveness of deep learning in financial applications.
Authors: Yunhua Pei, John Cartlidge, Anandadeep Mandal, Daniel Gold, Enrique Marcilio, Riccardo Mazzon
Abstract: Accurate financial market forecasting requires diverse data sources, including historical price trends, macroeconomic indicators, and financial news, each contributing unique predictive signals. However, existing methods often process these modalities independently or fail to effectively model their interactions. In this paper, we introduce Cross-Modal Temporal Fusion (CMTF), a novel transformer-based framework that integrates heterogeneous financial data to improve predictive accuracy. Our approach employs attention mechanisms to dynamically weight the contribution of different modalities, along with a specialized tensor interpretation module for feature extraction. To facilitate rapid model iteration in industry applications, we incorporate a mature auto-training scheme that streamlines optimization. When applied to real-world financial datasets, CMTF demonstrates improvements over baseline models in forecasting stock price movements and provides a scalable and effective solution for cross-modal integration in financial market prediction.
Authors: Zinuo You, John Cartlidge, Karen Elliott, Menghan Ge, Daniel Gold
Abstract: Existing portfolio management approaches are often black-box models due to safety and commercial issues in the industry. However, their performance can vary considerably whenever market conditions or internal trading strategies change. Furthermore, evaluating these non-transparent systems is expensive, where certain budgets limit observations of the systems. Therefore, optimizing performance while controlling the potential risk of these financial systems has become a critical challenge. This work presents a novel Bayesian optimization framework to optimize black-box portfolio management models under limited observations. In conventional Bayesian optimization settings, the objective function is to maximize the expectation of performance metrics. However, simply maximizing performance expectations leads to erratic optimization trajectories, which exacerbate risk accumulation in portfolio management. Meanwhile, this can lead to misalignment between the target distribution and the actual distribution of the black-box model. To mitigate this problem, we propose an adaptive weight Lagrangian estimator considering dual objective, which incorporates maximizing model performance and minimizing variance of model observations. Extensive experiments demonstrate the superiority of our approach over five backtest settings with three black-box stock portfolio management models. Ablation studies further verify the effectiveness of the proposed estimator.
Authors: Nikolay Manchev, Luis C. Garcia-Peraza-Herrera
Abstract: Recurrent Neural Networks (RNNs) are commonly used for real-time processing, streaming data, and cases where the amount of training samples is limited. Backpropagation Through Time (BPTT) is the predominant algorithm for training RNNs; however, it is frequently criticized for being prone to exploding and vanishing gradients and being biologically implausible. In this paper, we present and evaluate a target propagation-based method for RNNs, which uses local updates and seeks to reduce the said instabilities. Having stable RNN models increases their practical use in a wide range of fields such as natural language processing, time-series forecasting, anomaly detection, control systems, and robotics. The proposed solution uses local representation alignment (LRA). We thoroughly analyze the performance of this method, experiment with normalization and different local error functions, and invalidate certain assumptions about the behavior of this type of learning. Namely, we demonstrate that despite the decomposition of the network into sub-graphs, the model still suffers from vanishing gradients. We also show that gradient clipping as proposed in LRA has little to no effect on network performance. This results in an LRA RNN model that is very difficult to train due to vanishing gradients. We address this by introducing gradient regularization in the direction of the update and demonstrate that this modification promotes gradient flow and meaningfully impacts convergence. We compare and discuss the performance of the algorithm, and we show that the regularized LRA RNN considerably outperforms the unregularized version on three landmark tasks: temporal order, 3-bit temporal order, and random permutation.
Authors: Armin Iske, Lennart Ohlsen
Abstract: We develop sampling formulas for high-dimensional functions in reproducing kernel Hilbert spaces, where we rely on irregular samples that are taken at determining sequences of data points. We place particular emphasis on sampling formulas for tensor product kernels, where we show that determining irregular samples in lower dimensions can be composed to obtain a tensor of determining irregular samples in higher dimensions. This in turn reduces the computational complexity of sampling formulas for high-dimensional functions quite significantly.
Authors: Yuling Jiao, Yanming Lai, Yang Wang, Bokai Yan
Abstract: The Transformer model is widely used in various application areas of machine learning, such as natural language processing. This paper investigates the approximation of the H\"older continuous function class $\mathcal{H}_{Q}^{\beta}\left([0,1]^{d\times n},\mathbb{R}^{d\times n}\right)$ by Transformers and constructs several Transformers that can overcome the curse of dimensionality. These Transformers consist of one self-attention layer with one head and the softmax function as the activation function, along with several feedforward layers. For example, to achieve an approximation accuracy of $\epsilon$, if the activation functions of the feedforward layers in the Transformer are ReLU and floor, only $\mathcal{O}\left(\log\frac{1}{\epsilon}\right)$ layers of feedforward layers are needed, with widths of these layers not exceeding $\mathcal{O}\left(\frac{1}{\epsilon^{2/\beta}}\log\frac{1}{\epsilon}\right)$. If other activation functions are allowed in the feedforward layers, the width of the feedforward layers can be further reduced to a constant. These results demonstrate that Transformers have a strong expressive capability. The construction in this paper is based on the Kolmogorov-Arnold Representation Theorem and does not require the concept of contextual mapping, hence our proof is more intuitively clear compared to previous Transformer approximation works. Additionally, the translation technique proposed in this paper helps to apply the previous approximation results of feedforward neural networks to Transformer research.
Authors: Djohan Bonnet, Kellian Cottart, Tifenn Hirtzlin, Tarcisius Januel, Thomas Dalgaty, Elisa Vianello, Damien Querlioz
Abstract: Biological synapses effortlessly balance memory retention and flexibility, yet artificial neural networks still struggle with the extremes of catastrophic forgetting and catastrophic remembering. Here, we introduce Metaplasticity from Synaptic Uncertainty (MESU), a Bayesian framework that updates network parameters according their uncertainty. This approach allows a principled combination of learning and forgetting that ensures that critical knowledge is preserved while unused or outdated information is gradually released. Unlike standard Bayesian approaches -- which risk becoming overly constrained, and popular continual-learning methods that rely on explicit task boundaries, MESU seamlessly adapts to streaming data. It further provides reliable epistemic uncertainty estimates, allowing out-of-distribution detection, the only computational cost being to sample the weights multiple times to provide proper output statistics. Experiments on image-classification benchmarks demonstrate that MESU mitigates catastrophic forgetting, while maintaining plasticity for new tasks. When training 200 sequential permuted MNIST tasks, MESU outperforms established continual learning techniques in terms of accuracy, capability to learn additional tasks, and out-of-distribution data detection. Additionally, due to its non-reliance on task boundaries, MESU outperforms conventional learning techniques on the incremental training of CIFAR-100 tasks consistently in a wide range of scenarios. Our results unify ideas from metaplasticity, Bayesian inference, and Hessian-based regularization, offering a biologically-inspired pathway to robust, perpetual learning.
Authors: Zhenkai Qin, Feng Zhu, Huan Zeng, Xunyi Nong
Abstract: The demand for lightweight models in image classification tasks under resource-constrained environments necessitates a balance between computational efficiency and robust feature representation. Traditional attention mechanisms, despite their strong feature modeling capability, often struggle with high computational complexity and structural rigidity, limiting their applicability in scenarios with limited computational resources (e.g., edge devices or real-time systems). To address this, we propose the Multi-Agent Aggregation Module (MAAM), a lightweight attention architecture integrated with the MindSpore framework. MAAM employs three parallel agent branches with independently parameterized operations to extract heterogeneous features, adaptively fused via learnable scalar weights, and refined through a convolutional compression layer. Leveraging MindSpore's dynamic computational graph and operator fusion, MAAM achieves 87.0% accuracy on the CIFAR-10 dataset, significantly outperforming conventional CNN (58.3%) and MLP (49.6%) models, while improving training efficiency by 30%. Ablation studies confirm the critical role of agent attention (accuracy drops to 32.0% if removed) and compression modules (25.5% if omitted), validating their necessity for maintaining discriminative feature learning. The framework's hardware acceleration capabilities and minimal memory footprint further demonstrate its practicality, offering a deployable solution for image classification in resource-constrained scenarios without compromising accuracy.
Authors: Weiqi Qin, Yuxin Liu, Dongze Wu, Zhenkai Qin, Qining Luo
Abstract: Aiming at the problems of low accuracy and large error fluctuation of traditional traffic flow predictionmodels when dealing with multi-scale temporal features and dynamic change patterns. this paperproposes a multi-scale time series information modelling model MSTIM based on the Mindspore framework, which integrates long and short-term memory networks (LSTMs), convolutional neural networks (CNN), and the attention mechanism to improve the modelling accuracy and stability. The Metropolitan Interstate Traffic Volume (MITV) dataset was used for the experiments and compared and analysed with typical LSTM-attention models, CNN-attention models and LSTM-CNN models. The experimental results show that the MSTIM model achieves better results in the metrics of Mean Absolute Error (MAE), Mean Square Error (MSE), and Root Mean Square Error (RMSE), which significantly improves the accuracy and stability of the traffic volume prediction.
Authors: Jinghan Yang, Anupam Pani, Yunchao Zhang
Abstract: In the era of large-scale model training, the extensive use of available datasets has resulted in significant computational inefficiencies. To tackle this issue, we explore methods for identifying informative subsets of training data that can achieve comparable or even superior model performance. We propose a technique based on influence functions to determine which training samples should be included in the training set. We conducted empirical evaluations of our method on binary classification tasks utilizing logistic regression models. Our approach demonstrates performance comparable to that of training on the entire dataset while using only 10% of the data. Furthermore, we found that our method achieved even higher accuracy when trained with just 60% of the data.
Authors: Charalampos Kleitsikas, Nikolaos Korfiatis, Stefanos Leonardos, Carmine Ventre
Abstract: Cryptocurrency blockchains, beyond their primary role as distributed payment systems, are increasingly used to store and share arbitrary content, such as text messages and files. Although often non-financial, this hidden content can impact price movements by conveying private information, shaping sentiment, and influencing public opinion. However, current analyses of such data are limited in scope and scalability, primarily relying on manual classification or hand-crafted heuristics. In this work, we address these limitations by employing Natural Language Processing techniques to analyze, detect patterns, and extract public sentiment encoded within blockchain transactional data. Using a variety of Machine Learning techniques, we showcase for the first time the predictive power of blockchain-embedded sentiment in forecasting cryptocurrency price movements on the Bitcoin and Ethereum blockchains. Our findings shed light on a previously underexplored source of freely available, transparent, and immutable data and introduce blockchain sentiment analysis as a novel and robust framework for enhancing financial predictions in cryptocurrency markets. Incidentally, we discover an asymmetry between cryptocurrencies; Bitcoin has an informational advantage over Ethereum in that the sentiment embedded into transactional data is sufficient to predict its price movement.
Authors: Khoa Tran, Simon S. Woo
Abstract: Machine unlearning poses the challenge of ``how to eliminate the influence of specific data from a pretrained model'' in regard to privacy concerns. While prior research on approximated unlearning has demonstrated accuracy and efficiency in time complexity, we claim that it falls short of achieving exact unlearning, and we are the first to focus on fairness and robustness in machine unlearning algorithms. Our study presents fairness Conjectures for a well-trained model, based on the variance-bias trade-off characteristic, and considers their relevance to robustness. Our Conjectures are supported by experiments conducted on the two most widely used model architectures, ResNet and ViT, demonstrating the correlation between fairness and robustness: \textit{the higher fairness-gap is, the more the model is sensitive and vulnerable}. In addition, our experiments demonstrate the vulnerability of current state-of-the-art approximated unlearning algorithms to adversarial attacks, where their unlearned models suffer a significant drop in accuracy compared to the exact-unlearned models. We claim that our fairness-gap measurement and robustness metric should be used to evaluate the unlearning algorithm. Furthermore, we demonstrate that unlearning in the intermediate and last layers is sufficient and cost-effective for time and memory complexity.
Authors: Dejan Stancevic, Luca Ambrogioni
Abstract: The practical performance of generative diffusion models depends on the appropriate choice of the noise scheduling function, which can also be equivalently expressed as a time reparameterization. In this paper, we present a time scheduler that selects sampling points based on entropy rather than uniform time spacing, ensuring that each point contributes an equal amount of information to the final generation. We prove that this time reparameterization does not depend on the initial choice of time. Furthermore, we provide a tractable exact formula to estimate this \emph{entropic time} for a trained model using the training loss without substantial overhead. Alongside the entropic time, inspired by the optimality results, we introduce a rescaled entropic time. In our experiments with mixtures of Gaussian distributions and ImageNet, we show that using the (rescaled) entropic times greatly improves the inference performance of trained models. In particular, we found that the image quality in pretrained EDM2 models, as evaluated by FID and FD-DINO scores, can be substantially increased by the rescaled entropic time reparameterization without increasing the number of function evaluations, with greater improvements in the few NFEs regime.
Authors: Samuel Wertz, Arnaud Vandaele, Nicolas Gillis
Abstract: The Hadamard decomposition is a powerful technique for data analysis and matrix compression, which decomposes a given matrix into the element-wise product of two or more low-rank matrices. In this paper, we develop an efficient algorithm to solve this problem, leveraging an alternating optimization approach that decomposes the global non-convex problem into a series of convex sub-problems. To improve performance, we explore advanced initialization strategies inspired by the singular value decomposition (SVD) and incorporate acceleration techniques by introducing momentum-based updates. Beyond optimizing the two-matrix case, we also extend the Hadamard decomposition framework to support more than two low-rank matrices, enabling approximations with higher effective ranks while preserving computational efficiency. Finally, we conduct extensive experiments to compare our method with the existing gradient descent-based approaches for the Hadamard decomposition and with traditional low-rank approximation techniques. The results highlight the effectiveness of our proposed method across diverse datasets.
Authors: Jinhui Pang, Changqing Lin, Hao Lin, Jinglin He, Zhengjun Li, Zhihui Zhang, Xiaoshuai Hao
Abstract: Graph Few-Shot Class-Incremental Learning (GFSCIL) enables models to continually learn from limited samples of novel tasks after initial training on a large base dataset. Existing GFSCIL approaches typically utilize Prototypical Networks (PNs) for metric-based class representations and fine-tune the model during the incremental learning stage. However, these PN-based methods oversimplify learning via novel query set fine-tuning and fail to integrate Graph Continual Learning (GCL) techniques due to architectural constraints. To address these challenges, we propose a more rigorous and practical setting for GFSCIL that excludes query sets during the incremental training phase. Building on this foundation, we introduce Model-Agnostic Meta Graph Continual Learning (MEGA), aimed at effectively alleviating catastrophic forgetting for GFSCIL. Specifically, by calculating the incremental second-order gradient during the meta-training stage, we endow the model to learn high-quality priors that enhance incremental learning by aligning its behaviors across both the meta-training and incremental learning stages. Extensive experiments on four mainstream graph datasets demonstrate that MEGA achieves state-of-the-art results and enhances the effectiveness of various GCL methods in GFSCIL. We believe that our proposed MEGA serves as a model-agnostic GFSCIL paradigm, paving the way for future research.
Authors: Yichen Liu
Abstract: Heterogeneous treatment effect estimation in high-stakes applications demands models that simultaneously optimize precision, interpretability, and calibration. Many existing tree-based causal inference techniques, however, exhibit high estimation errors when applied to observational data because they struggle to capture complex interactions among factors and rely on static regularization schemes. In this work, we propose Dynamic Regularized Causal Boosted Decision Trees (CBDT), a novel framework that integrates variance regularization and average treatment effect calibration into the loss function of gradient boosted decision trees. Our approach dynamically updates the regularization parameters using gradient statistics to better balance the bias-variance tradeoff. Extensive experiments on standard benchmark datasets and real-world clinical data demonstrate that the proposed method significantly improves estimation accuracy while maintaining reliable coverage of true treatment effects. In an intensive care unit patient triage study, the method successfully identified clinically actionable rules and achieved high accuracy in treatment effect estimation. The results validate that dynamic regularization can effectively tighten error bounds and enhance both predictive performance and model interpretability.
Authors: Benjamin Cohen-Wang, Yung-Sung Chuang, Aleksander Madry
Abstract: Given a sequence of tokens generated by a language model, we may want to identify the preceding tokens that influence the model to generate this sequence. Performing such token attribution is expensive; a common approach is to ablate preceding tokens and directly measure their effects. To reduce the cost of token attribution, we revisit attention weights as a heuristic for how a language model uses previous tokens. Naive approaches to attribute model behavior with attention (e.g., averaging attention weights across attention heads to estimate a token's influence) have been found to be unreliable. To attain faithful attributions, we propose treating the attention weights of different attention heads as features. This way, we can learn how to effectively leverage attention weights for attribution (using signal from ablations). Our resulting method, Attribution with Attention (AT2), reliably performs on par with approaches that involve many ablations, while being significantly more efficient. To showcase the utility of AT2, we use it to prune less important parts of a provided context in a question answering setting, improving answer quality. We provide code for AT2 at https://github.com/MadryLab/AT2 .
Authors: Amin Noroozi, Sidratul Muntaha Esha, Mansoureh Ghari
Abstract: Childhood vaccination is a cornerstone of public health, yet disparities in vaccination coverage persist across England. These disparities are shaped by complex interactions among various factors, including geographic, demographic, socioeconomic, and cultural (GDSC) factors. Previous studies mostly rely on cross-sectional data and traditional statistical approaches that assess individual or limited sets of variables in isolation. Such methods may fall short in capturing the dynamic and multivariate nature of vaccine uptake. In this paper, we conducted a longitudinal machine learning analysis of childhood vaccination coverage across 150 districts in England from 2021 to 2024. Using vaccination data from NHS records, we applied hierarchical clustering to group districts by vaccination coverage into low- and high-coverage clusters. A CatBoost classifier was then trained to predict districts' vaccination clusters using their GDSC data. Finally, the SHapley Additive exPlanations (SHAP) method was used to interpret the predictors' importance. The classifier achieved high accuracies of 92.1, 90.6, and 86.3 in predicting districts' vaccination clusters for the years 2021-2022, 2022-2023, and 2023-2024, respectively. SHAP revealed that geographic, cultural, and demographic variables, particularly rurality, English language proficiency, the percentage of foreign-born residents, and ethnic composition, were the most influential predictors of vaccination coverage, whereas socioeconomic variables, such as deprivation and employment, consistently showed lower importance, especially in 2023-2024. Surprisingly, rural districts were significantly more likely to have higher vaccination rates. Additionally, districts with lower vaccination coverage had higher populations whose first language was not English, who were born outside the UK, or who were from ethnic minority groups.
Authors: Dmitrii Kharlapenko, Stepan Shabalin, Fazl Barez, Arthur Conmy, Neel Nanda
Abstract: Sparse autoencoders (SAEs) are a popular tool for interpreting large language model activations, but their utility in addressing open questions in interpretability remains unclear. In this work, we demonstrate their effectiveness by using SAEs to deepen our understanding of the mechanism behind in-context learning (ICL). We identify abstract SAE features that (i) encode the model's knowledge of which task to execute and (ii) whose latent vectors causally induce the task zero-shot. This aligns with prior work showing that ICL is mediated by task vectors. We further demonstrate that these task vectors are well approximated by a sparse sum of SAE latents, including these task-execution features. To explore the ICL mechanism, we adapt the sparse feature circuits methodology of Marks et al. (2024) to work for the much larger Gemma-1 2B model, with 30 times as many parameters, and to the more complex task of ICL. Through circuit finding, we discover task-detecting features with corresponding SAE latents that activate earlier in the prompt, that detect when tasks have been performed. They are causally linked with task-execution features through the attention and MLP sublayers.
Authors: Vinay Sharma, R\'emi Tanguy Oddon, Pietro Tesini, Jens Ravesloot, Cees Taal, Olga Fink
Abstract: Accurate real-time modeling of multi-body dynamical systems is essential for enabling digital twin applications across industries. While many data-driven approaches aim to learn system dynamics, jointly predicting internal loads and system trajectories remains a key challenge. This dual prediction is especially important for fault detection and predictive maintenance, where internal loads-such as contact forces-act as early indicators of faults, reflecting wear or misalignment before affecting motion. These forces also serve as inputs to degradation models (e.g., crack growth), enabling damage prediction and remaining useful life estimation. We propose Equi-Euler GraphNet, a physics-informed graph neural network (GNN) that simultaneously predicts internal forces and global trajectories in multi-body systems. In this mesh-free framework, nodes represent system components and edges encode interactions. Equi-Euler GraphNet introduces two inductive biases: (1) an equivariant message-passing scheme, interpreting edge messages as interaction forces consistent under Euclidean transformations; and (2) a temporal-aware iterative node update mechanism, based on Euler integration, to capture influence of distant interactions over time. Tailored for cylindrical roller bearings, it decouples ring dynamics from constrained motion of rolling elements. Trained on high-fidelity multiphysics simulations, Equi-Euler GraphNet generalizes beyond the training distribution, accurately predicting loads and trajectories under unseen speeds, loads, and configurations. It outperforms state-of-the-art GNNs focused on trajectory prediction, delivering stable rollouts over thousands of time steps with minimal error accumulation. Achieving up to a 200x speedup over conventional solvers while maintaining comparable accuracy, it serves as an efficient reduced-order model for digital twins, design, and maintenance.
Authors: Tamim Al Mahmud, Najeeb Jebreel, Josep Domingo-Ferrer, David Sanchez
Abstract: Large language models (LLMs) have recently revolutionized language processing tasks but have also brought ethical and legal issues. LLMs have a tendency to memorize potentially private or copyrighted information present in the training data, which might then be delivered to end users at inference time. When this happens, a naive solution is to retrain the model from scratch after excluding the undesired data. Although this guarantees that the target data have been forgotten, it is also prohibitively expensive for LLMs. Approximate unlearning offers a more efficient alternative, as it consists of ex post modifications of the trained model itself to prevent undesirable results, but it lacks forgetting guarantees because it relies solely on empirical evidence. In this work, we present DP2Unlearning, a novel LLM unlearning framework that offers formal forgetting guarantees at a significantly lower cost than retraining from scratch on the data to be retained. DP2Unlearning involves training LLMs on textual data protected using {\epsilon}-differential privacy (DP), which later enables efficient unlearning with the guarantees against disclosure associated with the chosen {\epsilon}. Our experiments demonstrate that DP2Unlearning achieves similar model performance post-unlearning, compared to an LLM retraining from scratch on retained data -- the gold standard exact unlearning -- but at approximately half the unlearning cost. In addition, with a reasonable computational cost, it outperforms approximate unlearning methods at both preserving the utility of the model post-unlearning and effectively forgetting the targeted information.
Authors: Lorenz Kummer, Wilfried N. Gansterer, Nils M. Kriege
Abstract: We investigate the vulnerability of Graph Neural Networks (GNNs) to bit-flip attacks (BFAs) by introducing an analytical framework to study the influence of architectural features, graph properties, and their interaction. The expressivity of GNNs refers to their ability to distinguish non-isomorphic graphs and depends on the encoding of node neighborhoods. We examine the vulnerability of neural multiset functions commonly used for this purpose and establish formal criteria to characterize a GNN's susceptibility to losing expressivity due to BFAs. This enables an analysis of the impact of homophily, graph structural variety, feature encoding, and activation functions on GNN robustness. We derive theoretical bounds for the number of bit flips required to degrade GNN expressivity on a dataset, identifying ReLU-activated GNNs operating on highly homophilous graphs with low-dimensional or one-hot encoded features as particularly susceptible. Empirical results using ten real-world datasets confirm the statistical significance of our key theoretical insights and offer actionable results to mitigate BFA risks in expressivity-critical applications.
Authors: Helen Jin, Anton Xue, Weiqiu You, Surbhi Goel, Eric Wong
Abstract: Stability guarantees are an emerging tool for evaluating feature attributions, but existing certification methods rely on smoothed classifiers and often yield conservative guarantees. To address these limitations, we introduce soft stability and propose a simple, model-agnostic, and sample-efficient stability certification algorithm (SCA) that provides non-trivial and interpretable guarantees for any attribution. Moreover, we show that mild smoothing enables a graceful tradeoff between accuracy and stability, in contrast to prior certification methods that require a more aggressive compromise. Using Boolean function analysis, we give a novel characterization of stability under smoothing. We evaluate SCA on vision and language tasks, and demonstrate the effectiveness of soft stability in measuring the robustness of explanation methods.
Authors: Weizhi Lu, Mingrui Chen, Weiyu Li
Abstract: In machine learning, quantization is widely used to simplify data representation and facilitate algorithm deployment on hardware. Given the fundamental role of classification in machine learning, it is crucial to investigate the impact of quantization on classification. Current research primarily focuses on quantization errors, operating under the premise that higher quantization errors generally result in lower classification performance. However, this premise lacks a solid theoretical foundation and often contradicts empirical findings. For instance, certain extremely low bit-width quantization methods, such as $\{0,1\}$-binary quantization and $\{0, \pm1\}$-ternary quantization, can achieve comparable or even superior classification accuracy compared to the original non-quantized data, despite exhibiting high quantization errors. To more accurately evaluate classification performance, we propose to directly investigate the feature discrimination of quantized data, instead of analyzing its quantization error. Interestingly, it is found that both binary and ternary quantization methods can improve, rather than degrade, the feature discrimination of the original data. This remarkable performance is validated through classification experiments across various data types, including images, speech, and texts.
Authors: Yu Wang, Shujie Liu, Shuai Lv, Gengshuo Liu
Abstract: Predicting the remaining useful life (RUL) of rotating machinery is critical for industrial safety and maintenance, but existing methods struggle with scarce target-domain data and unclear degradation dynamics. We propose a Meta-Learning and Knowledge Discovery-based Physics-Informed Neural Network (MKDPINN) to address these challenges. The method first maps noisy sensor data to a low-dimensional hidden state space via a Hidden State Mapper (HSM). A Physics-Guided Regulator (PGR) then learns unknown nonlinear PDEs governing degradation evolution, embedding these physical constraints into the PINN framework. This integrates data-driven and physics-based approaches. The framework uses meta-learning, optimizing across source-domain meta-tasks to enable few-shot adaptation to new target tasks. Experiments on industrial data and the C-MAPSS benchmark show MKDPINN outperforms baselines in generalization and accuracy, proving its effectiveness for RUL prediction under data scarcity
Authors: Nguyen Kim Hai Bui, Nguyen Duy Chien, P\'eter Kov\'acs, Gerg\H{o} Bogn\'ar
Abstract: Financial prediction is a complex and challenging task of time series analysis and signal processing, expected to model both short-term fluctuations and long-term temporal dependencies. Transformers have remarkable success mostly in natural language processing using attention mechanism, which also influenced the time series community. The ability to capture both short and long-range dependencies helps to understand the financial market and to recognize price patterns, leading to successful applications of Transformers in stock prediction. Although, the previous research predominantly focuses on individual features and singular predictions, that limits the model's ability to understand broader market trends. In reality, within sectors such as finance and technology, companies belonging to the same industry often exhibit correlated stock price movements. In this paper, we develop a novel neural network architecture by integrating Time2Vec with the Encoder of the Transformer model. Based on the study of different markets, we propose a novel correlation feature selection method. Through a comprehensive fine-tuning of multiple hyperparameters, we conduct a comparative analysis of our results against benchmark models. We conclude that our method outperforms other state-of-the-art encoding methods such as positional encoding, and we also conclude that selecting correlation features enhance the accuracy of predicting multiple stock prices.
Authors: Yixuan Even Xu, Yash Savani, Fei Fang, Zico Kolter
Abstract: Reinforcement learning (RL) has emerged as a powerful paradigm for enhancing reasoning capabilities in large language models, but faces a fundamental asymmetry in computation and memory requirements: inference is embarrassingly parallel with a minimal memory footprint, while policy updates require extensive synchronization and are memory-intensive. To address this asymmetry, we introduce PODS (Policy Optimization with Down-Sampling), a framework that strategically decouples these phases by generating numerous rollouts in parallel but updating only on an informative subset. Within this framework, we develop max-variance down-sampling, a theoretically motivated method that selects rollouts with maximally diverse reward signals. We prove that this approach has an efficient algorithmic solution, and empirically demonstrate that GRPO with PODS using max-variance down-sampling achieves superior performance over standard GRPO on the GSM8K benchmark.
Authors: Eric Nuertey Coleman, Luigi Quarantiello, Ziyue Liu, Qinwen Yang, Samrat Mukherjee, Julio Hurtado, Vincenzo Lomonaco
Abstract: The emergence of large pre-trained networks has revolutionized the AI field, unlocking new possibilities and achieving unprecedented performance. However, these models inherit a fundamental limitation from traditional Machine Learning approaches: their strong dependence on the \textit{i.i.d.} assumption hinders their adaptability to dynamic learning scenarios. We believe the next breakthrough in AI lies in enabling efficient adaptation to evolving environments -- such as the real world -- where new data and tasks arrive sequentially. This challenge defines the field of Continual Learning (CL), a Machine Learning paradigm focused on developing lifelong learning neural models. One alternative to efficiently adapt these large-scale models is known Parameter-Efficient Fine-Tuning (PEFT). These methods tackle the issue of adapting the model to a particular data or scenario by performing small and efficient modifications, achieving similar performance to full fine-tuning. However, these techniques still lack the ability to adjust the model to multiple tasks continually, as they suffer from the issue of Catastrophic Forgetting. In this survey, we first provide an overview of CL algorithms and PEFT methods before reviewing the state-of-the-art on Parameter-Efficient Continual Fine-Tuning (PECFT). We examine various approaches, discuss evaluation metrics, and explore potential future research directions. Our goal is to highlight the synergy between CL and Parameter-Efficient Fine-Tuning, guide researchers in this field, and pave the way for novel future research directions.
Authors: Yassine Habchi, Hamza Kheddar, Yassine Himeur, Adel Belouchrani, Erchin Serpedin, Fouad Khelifi, Muhammad E. H. Chowdhury
Abstract: The rapid advancement of deep learning (DL) has transformed healthcare, particularly in cancer detection and diagnosis. DL surpasses traditional machine learning and human accuracy, making it a critical tool for identifying diseases. Despite numerous reviews on DL in healthcare, a comprehensive analysis of its role in cancer detection remains limited. Existing studies focus on specific aspects, leaving gaps in understanding its broader impact. This paper addresses these gaps by reviewing advanced DL techniques, including transfer learning (TL), reinforcement learning (RL), federated learning (FL), Transformers, and large language models (LLMs). These approaches enhance accuracy, tackle data scarcity, and enable decentralized learning while maintaining data privacy. TL adapts pre-trained models to new datasets, improving performance with limited labeled data. RL optimizes diagnostic pathways and treatment strategies, while FL fosters collaborative model development without sharing sensitive data. Transformers and LLMs, traditionally used in natural language processing, are now applied to medical data for improved interpretability. Additionally, this review examines these techniques' efficiency in cancer diagnosis, addresses challenges like data imbalance, and proposes solutions. It serves as a resource for researchers and practitioners, providing insights into current trends and guiding future research in advanced DL for cancer detection.
Authors: Hong Yang
Abstract: For a single-gateway LoRaWAN network, this study proposed a history-enhanced two-phase actor-critic algorithm with a shared transformer algorithm (HEAT) to improve network performance. HEAT considers uplink parameters and often neglected downlink parameters, and effectively integrates offline and online reinforcement learning, using historical data and real-time interaction to improve model performance. In addition, this study developed an open source LoRaWAN network simulator LoRaWANSim. The simulator considers the demodulator lock effect and supports multi-channel, multi-demodulator and bidirectional communication. Simulation experiments show that compared with the best results of all compared algorithms, HEAT improves the packet success rate and energy efficiency by 15% and 95%, respectively.
Authors: Jirui Yang, Zheyu Lin, Shuhan Yang, Zhihui Lu, Xin Du
Abstract: Embodied Intelligence (EI) systems integrated with large language models (LLMs) face significant security risks, particularly from jailbreak attacks that manipulate models into generating harmful outputs or executing unsafe physical actions. Traditional defense strategies, such as input filtering and output monitoring, often introduce high computational overhead or interfere with task performance in real-time embodied scenarios. To address these challenges, we propose Concept Enhancement Engineering (CEE), a novel defense framework that leverages representation engineering to enhance the safety of embodied LLMs by dynamically steering their internal activations. CEE operates by (1) extracting multilingual safety patterns from model activations, (2) constructing control directions based on safety-aligned concept subspaces, and (3) applying subspace concept rotation to reinforce safe behavior during inference. Our experiments demonstrate that CEE effectively mitigates jailbreak attacks while maintaining task performance, outperforming existing defense methods in both robustness and efficiency. This work contributes a scalable and interpretable safety mechanism for embodied AI, bridging the gap between theoretical representation engineering and practical security applications. Our findings highlight the potential of latent-space interventions as a viable defense paradigm against emerging adversarial threats in physically grounded AI systems.
Authors: Salman Rahman, Liwei Jiang, James Shiffer, Genglin Liu, Sheriff Issaka, Md Rizwan Parvez, Hamid Palangi, Kai-Wei Chang, Yejin Choi, Saadia Gabriel
Abstract: Multi-turn interactions with language models (LMs) pose critical safety risks, as harmful intent can be strategically spread across exchanges. Yet, the vast majority of prior work has focused on single-turn safety, while adaptability and diversity remain among the key challenges of multi-turn red-teaming. To address these challenges, we present X-Teaming, a scalable framework that systematically explores how seemingly harmless interactions escalate into harmful outcomes and generates corresponding attack scenarios. X-Teaming employs collaborative agents for planning, attack optimization, and verification, achieving state-of-the-art multi-turn jailbreak effectiveness and diversity with success rates up to 98.1% across representative leading open-weight and closed-source models. In particular, X-Teaming achieves a 96.2% attack success rate against the latest Claude 3.7 Sonnet model, which has been considered nearly immune to single-turn attacks. Building on X-Teaming, we introduce XGuard-Train, an open-source multi-turn safety training dataset that is 20x larger than the previous best resource, comprising 30K interactive jailbreaks, designed to enable robust multi-turn safety alignment for LMs. Our work offers essential tools and insights for mitigating sophisticated conversational attacks, advancing the multi-turn safety of LMs.
Authors: Haomin Zuo, Zhengyang Li, Jiangchuan Gong, Zhen Tian
Abstract: As urbanization speeds up and traffic flow increases, the issue of pavement distress is becoming increasingly pronounced, posing a severe threat to road safety and service life. Traditional methods of pothole detection rely on manual inspection, which is not only inefficient but also costly. This paper proposes an intelligent road crack detection and analysis system, based on the enhanced YOLOv8 deep learning framework. A target segmentation model has been developed through the training of 4029 images, capable of efficiently and accurately recognizing and segmenting crack regions in roads. The model also analyzes the segmented regions to precisely calculate the maximum and minimum widths of cracks and their exact locations. Experimental results indicate that the incorporation of ECA and CBAM attention mechanisms substantially enhances the model's detection accuracy and efficiency, offering a novel solution for road maintenance and safety monitoring.
Authors: Maarten C. Vonk, Mauricio Gonzalez Soto, Anna V. Kononova
Abstract: Causality and game theory are two influential fields that contribute significantly to decision-making in various domains. Causality defines and models causal relationships in complex policy problems, while game theory provides insights into strategic interactions among stakeholders with competing interests. Integrating these frameworks has led to significant theoretical advancements with the potential to improve decision-making processes. However, practical applications of these developments remain underexplored. To support efforts toward implementation, this paper clarifies key concepts in game theory and causality that are essential to their intersection, particularly within the context of probabilistic graphical models. By rigorously examining these concepts and illustrating them with intuitive, consistent examples, we clarify the required inputs for implementing these models, provide practitioners with insights into their application and selection across different scenarios, and reference existing research that supports their implementation. We hope this work encourages broader adoption of these models in real-world scenarios.
Authors: Bokwang Hwang, Seonkyu Lim, Taewoong Kim, Yongjae Geun, Sunghyun Bang, Sohyun Park, Jihyun Park, Myeonggyu Lee, Jinwoo Lee, Yerin Kim, Jinsun Yoo, Jingyeong Hong, Jina Park, Yongchan Kim, Suhyun Kim, Younggyun Hahm, Yiseul Lee, Yejee Kang, Chanhyuk Yoon, Chansu Lee, Heeyewon Jeong, Jiyeon Lee, Seonhye Gu, Hyebin Kang, Yousang Cho, Hangyeol Yoo, KyungTae Lim
Abstract: We introduce KFinEval-Pilot, a benchmark suite specifically designed to evaluate large language models (LLMs) in the Korean financial domain. Addressing the limitations of existing English-centric benchmarks, KFinEval-Pilot comprises over 1,000 curated questions across three critical areas: financial knowledge, legal reasoning, and financial toxicity. The benchmark is constructed through a semi-automated pipeline that combines GPT-4-generated prompts with expert validation to ensure domain relevance and factual accuracy. We evaluate a range of representative LLMs and observe notable performance differences across models, with trade-offs between task accuracy and output safety across different model families. These results highlight persistent challenges in applying LLMs to high-stakes financial applications, particularly in reasoning and safety. Grounded in real-world financial use cases and aligned with the Korean regulatory and linguistic context, KFinEval-Pilot serves as an early diagnostic tool for developing safer and more reliable financial AI systems.
Authors: Ummay Maria Muna, Md. Mehedi Hasan Shawon, Md Jobayer, Sumaiya Akter, Saifur Rahman Sabuj
Abstract: Brain-computer interfaces (BCI) in electroencephalography (EEG)-based motor imagery classification offer promising solutions in neurorehabilitation and assistive technologies by enabling communication between the brain and external devices. However, the non-stationary nature of EEG signals and significant inter-subject variability cause substantial challenges for developing robust cross-subject classification models. This paper introduces a novel Spatial-Spectral-Temporal Attention Fusion (SSTAF) Transformer specifically designed for upper-limb motor imagery classification. Our architecture consists of a spectral transformer and a spatial transformer, followed by a transformer block and a classifier network. Each module is integrated with attention mechanisms that dynamically attend to the most discriminative patterns across multiple domains, such as spectral frequencies, spatial electrode locations, and temporal dynamics. The short-time Fourier transform is incorporated to extract features in the time-frequency domain to make it easier for the model to obtain a better feature distinction. We evaluated our SSTAF Transformer model on two publicly available datasets, the EEGMMIDB dataset, and BCI Competition IV-2a. SSTAF Transformer achieves an accuracy of 76.83% and 68.30% in the data sets, respectively, outperforms traditional CNN-based architectures and a few existing transformer-based approaches.
Authors: Weijie Shi, Jipeng Zhang, Yaguang Wu, Jingzhi Fang, Ruiyuan Zhang, Jiajie Xu, Jia Zhu, Hao Chen, Yao Zhao, Sirui Han, Xiaofang Zhou
Abstract: Large language models (LLMs) are commonly trained on multi-domain datasets, where domain sampling strategies significantly impact model performance due to varying domain importance across downstream tasks. Existing approaches for optimizing domain-level sampling strategies struggle with maintaining intra-domain consistency and accurately measuring domain impact. In this paper, we present Domain Impact-aware Data Sampling (DIDS). To ensure intra-domain consistency, a gradient clustering algorithm is proposed to group training data based on their learning effects, where a proxy language model and dimensionality reduction are employed to reduce computational overhead. To accurately measure domain impact, we develop a Fisher Information Matrix (FIM) guided metric that quantifies how domain-specific parameter updates affect the model's output distributions on downstream tasks, with theoretical guarantees. Furthermore, to determine optimal sampling ratios, DIDS combines both the FIM-guided domain impact assessment and loss learning trajectories that indicate domain-specific potential, while accounting for diminishing marginal returns. Extensive experiments demonstrate that DIDS achieves 3.4% higher average performance while maintaining comparable training efficiency.
Authors: Sayed Pouria Talebi, Clive Cheong Took, Danilo P. Mandic
Abstract: This article considers the problem of designing adaption and optimisation techniques for training quantum learning machines. To this end, the division algebra of quaternions is used to derive an effective model for representing computation and measurement operations on qubits. In turn, the derived model, serves as the foundation for formulating an adaptive learning problem on principal quantum learning units, thereby establishing quantum information processing units akin to that of neurons in classical approaches. Then, leveraging the modern HR-calculus, a comprehensive training framework for learning on quantum machines is developed. The quaternion-valued model accommodates mathematical tractability and establishment of performance criteria, such as convergence conditions.
Authors: Jiang-Xin Shi, Tong Wei, Yu-Feng Li
Abstract: The fine-tuning paradigm has emerged as a prominent approach for addressing long-tail learning tasks in the era of foundation models. However, the impact of fine-tuning strategies on long-tail learning performance remains unexplored. In this work, we disclose that existing paradigms exhibit a profound misuse of fine-tuning methods, leaving significant room for improvement in both efficiency and accuracy. Specifically, we reveal that heavy fine-tuning (fine-tuning a large proportion of model parameters) can lead to non-negligible performance deterioration on tail classes, whereas lightweight fine-tuning demonstrates superior effectiveness. Through comprehensive theoretical and empirical validation, we identify this phenomenon as stemming from inconsistent class conditional distributions induced by heavy fine-tuning. Building on this insight, we propose LIFT+, an innovative lightweight fine-tuning framework to optimize consistent class conditions. Furthermore, LIFT+ incorporates semantic-aware initialization, minimalist data augmentation, and test-time ensembling to enhance adaptation and generalization of foundation models. Our framework provides an efficient and accurate pipeline that facilitates fast convergence and model compactness. Extensive experiments demonstrate that LIFT+ significantly reduces both training epochs (from $\sim$100 to $\leq$15) and learned parameters (less than 1%), while surpassing state-of-the-art approaches by a considerable margin. The source code is available at https://github.com/shijxcs/LIFT-plus.
Authors: Robert Gruhlke, Matei Hanu, Claudia Schillings, Philipp Wacker
Abstract: We introduce a gradient-free framework for Bayesian Optimal Experimental Design (BOED) in sequential settings, aimed at complex systems where gradient information is unavailable. Our method combines Ensemble Kalman Inversion (EKI) for design optimization with the Affine-Invariant Langevin Dynamics (ALDI) sampler for efficient posterior sampling-both of which are derivative-free and ensemble-based. To address the computational challenges posed by nested expectations in BOED, we propose variational Gaussian and parametrized Laplace approximations that provide tractable upper and lower bounds on the Expected Information Gain (EIG). These approximations enable scalable utility estimation in high-dimensional spaces and PDE-constrained inverse problems. We demonstrate the performance of our framework through numerical experiments ranging from linear Gaussian models to PDE-based inference tasks, highlighting the method's robustness, accuracy, and efficiency in information-driven experimental design.
Authors: Ludovico T. Giorgini, Fabrizio Falasca, Andre N. Souza
Abstract: We present a novel data-driven framework for estimating the response of higher-order moments of nonlinear stochastic systems to small external perturbations. The classical Generalized Fluctuation-Dissipation Theorem (GFDT) links the unperturbed steady-state distribution to the system's linear response. Standard implementations rely on Gaussian approximations, which can often accurately predict the mean response but usually introduce significant biases in higher-order moments, such as variance, skewness, and kurtosis. To address this limitation, we combine GFDT with recent advances in score-based generative modeling, which enable direct estimation of the score function from data without requiring full density reconstruction. Our method is validated on three reduced-order stochastic models relevant to climate dynamics: a scalar stochastic model for low-frequency climate variability, a slow-fast triad model mimicking key features of the El Nino-Southern Oscillation (ENSO), and a six-dimensional stochastic barotropic model capturing atmospheric regime transitions. In all cases, the approach captures strongly nonlinear and non-Gaussian features of the system's response, outperforming traditional Gaussian approximations.
Authors: Lea Kunkel, Mathias Trabs
Abstract: Flow Matching has recently gained attention in generative modeling as a simple and flexible alternative to diffusion models, the current state of the art. While existing statistical guarantees adapt tools from the analysis of diffusion models, we take a different perspective by connecting Flow Matching to kernel density estimation. We first verify that the kernel density estimator matches the optimal rate of convergence in Wasserstein distance up to logarithmic factors, improving existing bounds for the Gaussian kernel. Based on this result, we prove that for sufficiently large networks, Flow Matching also achieves the optimal rate up to logarithmic factors, providing a theoretical foundation for the empirical success of this method. Finally, we provide a first justification of Flow Matching's effectiveness in high-dimensional settings by showing that rates improve when the target distribution lies on a lower-dimensional linear subspace.
Authors: Chen Wang, Fei Xia, Wenhao Yu, Tingnan Zhang, Ruohan Zhang, C. Karen Liu, Li Fei-Fei, Jie Tan, Jacky Liang
Abstract: Learning to perform manipulation tasks from human videos is a promising approach for teaching robots. However, many manipulation tasks require changing control parameters during task execution, such as force, which visual data alone cannot capture. In this work, we leverage sensing devices such as armbands that measure human muscle activities and microphones that record sound, to capture the details in the human manipulation process, and enable robots to extract task plans and control parameters to perform the same task. To achieve this, we introduce Chain-of-Modality (CoM), a prompting strategy that enables Vision Language Models to reason about multimodal human demonstration data -- videos coupled with muscle or audio signals. By progressively integrating information from each modality, CoM refines a task plan and generates detailed control parameters, enabling robots to perform manipulation tasks based on a single multimodal human video prompt. Our experiments show that CoM delivers a threefold improvement in accuracy for extracting task plans and control parameters compared to baselines, with strong generalization to new task setups and objects in real-world robot experiments. Videos and code are available at https://chain-of-modality.github.io
Authors: Long Li, Jiajia Li, Dong Chen, Lina Pu, Haibo Yao, Yanbo Huang
Abstract: In modern smart agriculture, object detection plays a crucial role by enabling automation, precision farming, and monitoring of resources. From identifying crop health and pest infestations to optimizing harvesting processes, accurate object detection enhances both productivity and sustainability. However, training object detection models often requires large-scale data collection and raises privacy concerns, particularly when sensitive agricultural data is distributed across farms. To address these challenges, we propose VLLFL, a vision-language model-based lightweight federated learning framework (VLLFL). It harnesses the generalization and context-aware detection capabilities of the vision-language model (VLM) and leverages the privacy-preserving nature of federated learning. By training a compact prompt generator to boost the performance of the VLM deployed across different farms, VLLFL preserves privacy while reducing communication overhead. Experimental results demonstrate that VLLFL achieves 14.53% improvement in the performance of VLM while reducing 99.3% communication overhead. Spanning tasks from identifying a wide variety of fruits to detecting harmful animals in agriculture, the proposed framework offers an efficient, scalable, and privacy-preserving solution specifically tailored to agricultural applications.
Authors: Racheal Mukisa, Arvind K. Bansal
Abstract: Automated noninvasive cardiac diagnosis plays a critical role in the early detection of cardiac disorders and cost-effective clinical management. Automated diagnosis involves the automated segmentation and analysis of cardiac images. Precise delineation of cardiac substructures and extraction of their morphological attributes are essential for evaluating the cardiac function, and diagnosing cardiovascular disease such as cardiomyopathy, valvular diseases, abnormalities related to septum perforations, and blood-flow rate. Semantic segmentation labels the CMR image at the pixel level, and localizes its subcomponents to facilitate the detection of abnormalities, including abnormalities in cardiac wall motion in an aging heart with muscle abnormalities, vascular abnormalities, and valvular abnormalities. In this paper, we describe a model to improve semantic segmentation of CMR images. The model extracts edge-attributes and context information during down-sampling of the U-Net and infuses this information during up-sampling to localize three major cardiac structures: left ventricle cavity (LV); right ventricle cavity (RV); and LV myocardium (LMyo). We present an algorithm and performance results. A comparison of our model with previous leading models, using similarity metrics between actual image and segmented image, shows that our approach improves Dice similarity coefficient (DSC) by 2%-11% and lowers Hausdorff distance (HD) by 1.6 to 5.7 mm.
Authors: Yu Gan, Mohadeseh Azar, Nitish Kumar Chandra, Xin Jin, Jinglei Cheng, Kaushik P. Seshadreesan, Junyu Liu
Abstract: The development of large-scale quantum communication networks faces critical challenges due to photon loss and decoherence in optical fiber channels. These fundamentally limit transmission distances and demand dense networks of repeater stations. This work investigates using vacuum beam guides (VBGs)-a promising ultra-low-loss transmission platform-as an alternative to traditional fiber links. By incorporating VBGs into repeater-based architectures, we demonstrate that the inter-repeater spacing can be substantially extended, resulting in fewer required nodes and significantly reducing hardware and operational complexity. We perform a cost-function analysis to quantify performance trade-offs across first, second, and third-generation repeaters. Our results show that first-generation repeaters reduce costs dramatically by eliminating entanglement purification. Third-generation repeaters benefit from improved link transmission success, which is crucial for quantum error correction. In contrast, second-generation repeaters exhibit a more nuanced response; although transmission loss is reduced, their performance remains primarily limited by logical gate errors rather than channel loss. These findings highlight that while all repeater generations benefit from reduced photon loss, the magnitude of improvement depends critically on the underlying error mechanisms. Vacuum beam guides thus emerge as a powerful enabler for scalable, high-performance quantum networks, particularly in conjunction with near-term quantum hardware capabilities.
Authors: Varij Saini, Rudraksh Gupta, Neel Soni
Abstract: This technical report presents a comprehensive analysis of malware classification using OpCode sequences. Two distinct approaches are evaluated: traditional machine learning using n-gram analysis with Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Decision Tree classifiers; and a deep learning approach employing a Convolutional Neural Network (CNN). The traditional machine learning approach establishes a baseline using handcrafted 1-gram and 2-gram features from disassembled malware samples. The deep learning methodology builds upon the work proposed in "Deep Android Malware Detection" by McLaughlin et al. and evaluates the performance of a CNN model trained to automatically extract features from raw OpCode data. Empirical results are compared using standard performance metrics (accuracy, precision, recall, and F1-score). While the SVM classifier outperforms other traditional techniques, the CNN model demonstrates competitive performance with the added benefit of automated feature extraction.
Authors: Samuel Audia, Soheil Feizi, Matthias Zwicker, Dinesh Manocha
Abstract: Neural networks that map between low dimensional spaces are ubiquitous in computer graphics and scientific computing; however, in their naive implementation, they are unable to learn high frequency information. We present a comprehensive analysis comparing the two most common techniques for mitigating this spectral bias: Fourier feature encodings (FFE) and multigrid parametric encodings (MPE). FFEs are seen as the standard for low dimensional mappings, but MPEs often outperform them and learn representations with higher resolution and finer detail. FFE's roots in the Fourier transform, make it susceptible to aliasing if pushed too far, while MPEs, which use a learned grid structure, have no such limitation. To understand the difference in performance, we use the neural tangent kernel (NTK) to evaluate these encodings through the lens of an analogous kernel regression. By finding a lower bound on the smallest eigenvalue of the NTK, we prove that MPEs improve a network's performance through the structure of their grid and not their learnable embedding. This mechanism is fundamentally different from FFEs, which rely solely on their embedding space to improve performance. Results are empirically validated on a 2D image regression task using images taken from 100 synonym sets of ImageNet and 3D implicit surface regression on objects from the Stanford graphics dataset. Using peak signal-to-noise ratio (PSNR) and multiscale structural similarity (MS-SSIM) to evaluate how well fine details are learned, we show that the MPE increases the minimum eigenvalue by 8 orders of magnitude over the baseline and 2 orders of magnitude over the FFE. The increase in spectrum corresponds to a 15 dB (PSNR) / 0.65 (MS-SSIM) increase over baseline and a 12 dB (PSNR) / 0.33 (MS-SSIM) increase over the FFE.
Authors: Hsin-Yi Lin, Huan-Hsin Tseng, Samuel Yen-Chi Chen, Shinjae Yoo
Abstract: Conventional Variational Quantum Circuits (VQCs) for Quantum Machine Learning typically rely on a fixed Hermitian observable, often built from Pauli operators. Inspired by the Heisenberg picture, we propose an adaptive non-local measurement framework that substantially increases the model complexity of the quantum circuits. Our introduction of dynamical Hermitian observables with evolving parameters shows that optimizing VQC rotations corresponds to tracing a trajectory in the observable space. This viewpoint reveals that standard VQCs are merely a special case of the Heisenberg representation. Furthermore, we show that properly incorporating variational rotations with non-local observables enhances qubit interaction and information mixture, admitting flexible circuit designs. Two non-local measurement schemes are introduced, and numerical simulations on classification tasks confirm that our approach outperforms conventional VQCs, yielding a more powerful and resource-efficient approach as a Quantum Neural Network.
Authors: Racheal Mukisa, Arvind K. Bansal
Abstract: We propose an enhanced deep learning-based model for image segmentation of the left and right ventricles and myocardium scar tissue from cardiac magnetic resonance (CMR) images. The proposed technique integrates UNet, channel and spatial attention, edge-detection based skip-connection and deep supervised learning to improve the accuracy of the CMR image-segmentation. Images are processed using multiple channels to generate multiple feature-maps. We built a dual attention-based model to integrate channel and spatial attention. The use of extracted edges in skip connection improves the reconstructed images from feature-maps. The use of deep supervision reduces vanishing gradient problems inherent in classification based on deep neural networks. The algorithms for dual attention-based model, corresponding implementation and performance results are described. The performance results show that this approach has attained high accuracy: 98% Dice Similarity Score (DSC) and significantly lower Hausdorff Distance (HD). The performance results outperform other leading techniques both in DSC and HD.
Authors: Juliette Bertrand, Sophie Giffard-Roisin, James Hollingsworth, Julien Mairal
Abstract: Dense ground displacement measurements are crucial for geological studies but are impractical to collect directly. Traditionally, displacement fields are estimated using patch matching on optical satellite images from different acquisition times. While deep learning-based optical flow models are promising, their adoption in ground deformation analysis is hindered by challenges such as the absence of real ground truth, the need for sub-pixel precision, and temporal variations due to geological or anthropogenic changes. In particular, we identify that deep learning models relying on explicit correlation layers struggle at estimating small displacements in real-world conditions. Instead, we propose a model that employs iterative refinements with explicit warping layers and a correlation-independent backbone, enabling sub-pixel precision. Additionally, a non-convex variant of Total Variation regularization preserves fault-line sharpness while maintaining smoothness elsewhere. Our model significantly outperforms widely used geophysics methods on semi-synthetic benchmarks and generalizes well to challenging real-world scenarios captured by both medium- and high-resolution sensors. Project page: https://jbertrand89.github.io/microflow/.
Authors: Xinchen Wang, Pengfei Gao, Chao Peng, Ruida Hu, Cuiyun Gao
Abstract: Large language models (LLMs) have demonstrated strong capabilities in code generation, underscoring the critical need for rigorous and comprehensive evaluation. Existing evaluation approaches fall into three categories, including human-centered, metric-based, and LLM-based. Considering that human-centered approaches are labour-intensive and metric-based ones overly rely on reference answers, LLM-based approaches are gaining increasing attention due to their stronger contextual understanding capabilities and superior efficiency. However, the performance of LLM-based approaches remains limited due to: (1) lack of multisource domain knowledge, and (2) insufficient comprehension of complex code. To mitigate the limitations, we propose CodeVisionary, the first LLM-based agent framework for evaluating LLMs in code generation. CodeVisionary consists of two stages: (1) Multiscore knowledge analysis stage, which aims to gather multisource and comprehensive domain knowledge by formulating and executing a stepwise evaluation plan. (2) Negotiation-based scoring stage, which involves multiple judges engaging in discussions to better comprehend the complex code and reach a consensus on the evaluation score. Extensive experiments demonstrate that CodeVisionary achieves the best performance for evaluating LLMs in code generation, outperforming the best baseline methods with average improvements of 0.202, 0.139, and 0.117 in Pearson, Spearman, and Kendall-Tau coefficients, respectively. Besides, CodeVisionary provides detailed evaluation reports, which assist developers in identifying shortcomings and making improvements. The resources of CodeVisionary are available at https://anonymous.4open.science/r/CodeVisionary.
Authors: Jiasheng Wu, Jingjing Zhang, Zheng Lin, Zhe Chen, Xiong Wang, Wenjun Zhu, Yue Gao
Abstract: Recently, the rapid development of LEO satellite networks spurs another widespread concern-data processing at satellites. However, achieving efficient computation at LEO satellites in highly dynamic satellite networks is challenging and remains an open problem when considering the constrained computation capability of LEO satellites. For the first time, we propose a novel distributed learning framework named SFL-LEO by combining Federated Learning (FL) with Split Learning (SL) to accommodate the high dynamics of LEO satellite networks and the constrained computation capability of LEO satellites by leveraging the periodical orbit traveling feature. The proposed scheme allows training locally by introducing an asynchronous training strategy, i.e., achieving local update when LEO satellites disconnect with the ground station, to provide much more training space and thus increase the training performance. Meanwhile, it aggregates client-side sub-models at the ground station and then distributes them to LEO satellites by borrowing the idea from the federated learning scheme. Experiment results driven by satellite-ground bandwidth measured in Starlink demonstrate that SFL-LEO provides a similar accuracy performance with the conventional SL scheme because it can perform local training even within the disconnection duration.
Authors: Miit Daga, Priyasha Mohanty, Ram Krishna, Swarna Priya RM
Abstract: This systematic review discusses the methodological approaches and statistical confirmations of cross-cultural adaptations of cognitive evaluation tools used with different populations. The review considers six seminal studies on the methodology of cultural adaptation in Europe, Asia, Africa, and South America. The results indicate that proper adaptations need holistic models with demographic changes, and education explained as much as 26.76% of the variance in MoCA-H scores. Cultural-linguistic factors explained 6.89% of the variance in European adaptations of MoCA-H; however, another study on adapted MMSE and BCSB among Brazilian Indigenous populations reported excellent diagnostic performance, with a sensitivity of 94.4% and specificity of 99.2%. There was 78.5% inter-rater agreement on the evaluation of cultural adaptation using the Manchester Translation Evaluation Checklist. A paramount message of the paper is that community feedback is necessary for culturally appropriate preparation, standardized translation protocols also must be included, along with robust statistical validation methodologies for developing cognitive assessment instruments. This review supplies evidence-based frameworks for the further adaptation of cognitive assessments in increasingly diverse global health settings.
Authors: Yipeng Sun, Linda-Sophie Schneider, Mingxuan Gu, Siyuan Mei, Chengze Ye, Fabian Wagner, Siming Bayer, Andreas Maier
Abstract: Effective denoising is crucial in low-dose CT to enhance subtle structures and low-contrast lesions while preventing diagnostic errors. Supervised methods struggle with limited paired datasets, and self-supervised approaches often require multiple noisy images and rely on deep networks like U-Net, offering little insight into the denoising mechanism. To address these challenges, we propose an interpretable self-supervised single-image denoising framework -- Filter2Noise (F2N). Our approach introduces an Attention-Guided Bilateral Filter that adapted to each noisy input through a lightweight module that predicts spatially varying filter parameters, which can be visualized and adjusted post-training for user-controlled denoising in specific regions of interest. To enable single-image training, we introduce a novel downsampling shuffle strategy with a new self-supervised loss function that extends the concept of Noise2Noise to a single image and addresses spatially correlated noise. On the Mayo Clinic 2016 low-dose CT dataset, F2N outperforms the leading self-supervised single-image method (ZS-N2N) by 4.59 dB PSNR while improving transparency, user control, and parametric efficiency. These features provide key advantages for medical applications that require precise and interpretable noise reduction. Our code is demonstrated at https://github.com/sypsyp97/Filter2Noise.git .
Authors: Corentin Larroche
Abstract: Foundation models have recently emerged as a new paradigm in machine learning (ML). These models are pre-trained on large and diverse datasets and can subsequently be applied to various downstream tasks with little or no retraining. This allows people without advanced ML expertise to build ML applications, accelerating innovation across many fields. However, the adoption of foundation models in cybersecurity is hindered by their inability to efficiently process data such as network traffic captures or binary executables. The recent introduction of graph foundation models (GFMs) could make a significant difference, as graphs are well-suited to representing these types of data. We study the usability of GFMs in cybersecurity through the lens of one specific use case, namely lateral movement detection. Using a pre-trained GFM, we build a detector that reaches state-of-the-art performance without requiring any training on domain-specific data. This case study thus provides compelling evidence of the potential of GFMs for cybersecurity.
Authors: Avaneesh Devkota, Rachmad Vidya Wicaksana Putra, Muhammad Shafique
Abstract: The ability to train intelligent autonomous agents (such as mobile robots) on multiple tasks is crucial for adapting to dynamic real-world environments. However, state-of-the-art reinforcement learning (RL) methods only excel in single-task settings, and still struggle to generalize across multiple tasks due to task interference. Moreover, real-world environments also demand the agents to have data stream processing capabilities. Toward this, a state-of-the-art work employs Spiking Neural Networks (SNNs) to improve multi-task learning by exploiting temporal information in data stream, while enabling lowpower/energy event-based operations. However, it relies on fixed context/task-switching intervals during its training, hence limiting the scalability and effectiveness of multi-task learning. To address these limitations, we propose SwitchMT, a novel adaptive task-switching methodology for RL-based multi-task learning in autonomous agents. Specifically, SwitchMT employs the following key ideas: (1) a Deep Spiking Q-Network with active dendrites and dueling structure, that utilizes task-specific context signals to create specialized sub-networks; and (2) an adaptive task-switching policy that leverages both rewards and internal dynamics of the network parameters. Experimental results demonstrate that SwitchMT achieves superior performance in multi-task learning compared to state-of-the-art methods. It achieves competitive scores in multiple Atari games (i.e., Pong: -8.8, Breakout: 5.6, and Enduro: 355.2) compared to the state-of-the-art, showing its better generalized learning capability. These results highlight the effectiveness of our SwitchMT methodology in addressing task interference while enabling multi-task learning automation through adaptive task switching, thereby paving the way for more efficient generalist agents with scalable multi-task learning capabilities.
Authors: Azmarah Rizvi, Navojith Thamindu, A. M. N. H. Adhikari, W. P. U. Senevirathna, Dharshana Kasthurirathna, Lakmini Abeywardhana
Abstract: Sentiment analysis is crucial for brand reputation management in the banking sector, where customer feedback spans English, Sinhala, Singlish, and code-mixed text. Existing models struggle with low-resource languages like Sinhala and lack interpretability for practical use. This research develops a hybrid aspect-based sentiment analysis framework that enhances multilingual capabilities with explainable outputs. Using cleaned banking customer reviews, we fine-tune XLM-RoBERTa for Sinhala and code-mixed text, integrate domain-specific lexicon correction, and employ BERT-base-uncased for English. The system classifies sentiment (positive, neutral, negative) with confidence scores, while SHAP and LIME improve interpretability by providing real-time sentiment explanations. Experimental results show that our approaches outperform traditional transformer-based classifiers, achieving 92.3 percent accuracy and an F1-score of 0.89 in English and 88.4 percent in Sinhala and code-mixed content. An explainability analysis reveals key sentiment drivers, improving trust and transparency. A user-friendly interface delivers aspect-wise sentiment insights, ensuring accessibility for businesses. This research contributes to robust, transparent sentiment analysis for financial applications by bridging gaps in multilingual, low-resource NLP and explainability.
Authors: Xin Tang, Qian Chen, Wenjie Weng, Chao Jin, Zhang Liu, Jiacheng Wang, Geng Sun, Xiaohuan Li, Dusit Niyato
Abstract: Artificial Intelligence (AI)-driven convolutional neural networks enhance rescue, inspection, and surveillance tasks performed by low-altitude uncrewed aerial vehicles (UAVs) and ground computing nodes (GCNs) in unknown environments. However, their high computational demands often exceed a single UAV's capacity, leading to system instability, further exacerbated by the limited and dynamic resources of GCNs. To address these challenges, this paper proposes a novel cooperation framework involving UAVs, ground-embedded robots (GERs), and high-altitude platforms (HAPs), which enable resource pooling through UAV-to-GER (U2G) and UAV-to-HAP (U2H) communications to provide computing services for UAV offloaded tasks. Specifically, we formulate the multi-objective optimization problem of task assignment and exploration optimization in UAVs as a dynamic long-term optimization problem. Our objective is to minimize task completion time and energy consumption while ensuring system stability over time. To achieve this, we first employ the Lyapunov optimization technique to transform the original problem, with stability constraints, into a per-slot deterministic problem. We then propose an algorithm named HG-MADDPG, which combines the Hungarian algorithm with a generative diffusion model (GDM)-based multi-agent deep deterministic policy gradient (MADDPG) approach. We first introduce the Hungarian algorithm as a method for exploration area selection, enhancing UAV efficiency in interacting with the environment. We then innovatively integrate the GDM and multi-agent deep deterministic policy gradient (MADDPG) to optimize task assignment decisions, such as task offloading and resource allocation. Simulation results demonstrate the effectiveness of the proposed approach, with significant improvements in task offloading efficiency, latency reduction, and system stability compared to baseline methods.
Authors: Zongyuan Chen, Yan Xia, Jiayuan Liu, Jijia Liu, Wenhao Tang, Jiayu Chen, Feng Gao, Longfei Ma, Hongen Liao, Yu Wang, Chao Yu, Boyu Zhang, Fei Xing
Abstract: Soft robots exhibit inherent compliance and safety, which makes them particularly suitable for applications requiring direct physical interaction with humans, such as surgical procedures. However, their nonlinear and hysteretic behavior, resulting from the properties of soft materials, presents substantial challenges for accurate modeling and control. In this study, we present a soft robotic system designed for surgical applications and propose a hysteresis-aware whole-body neural network model that accurately captures and predicts the soft robot's whole-body motion, including its hysteretic behavior. Building upon the high-precision dynamic model, we construct a highly parallel simulation environment for soft robot control and apply an on-policy reinforcement learning algorithm to efficiently train whole-body motion control strategies. Based on the trained control policy, we developed a soft robotic system for surgical applications and validated it through phantom-based laser ablation experiments in a physical environment. The results demonstrate that the hysteresis-aware modeling reduces the Mean Squared Error (MSE) by 84.95 percent compared to traditional modeling methods. The deployed control algorithm achieved a trajectory tracking error ranging from 0.126 to 0.250 mm on the real soft robot, highlighting its precision in real-world conditions. The proposed method showed strong performance in phantom-based surgical experiments and demonstrates its potential for complex scenarios, including future real-world clinical applications.
Authors: Lam Dinh, Sihem Cherrared, Xiaofeng Huang, Fabrice Guillemin
Abstract: Large Language Models (LLMs) are likely to play a key role in Intent-Based Networking (IBN) as they show remarkable performance in interpreting human language as well as code generation, enabling the translation of high-level intents expressed by humans into low-level network configurations. In this paper, we leverage closed-source language models (i.e., Google Gemini 1.5 pro, ChatGPT-4) and open-source models (i.e., LLama, Mistral) to investigate their capacity to generate E2E network configurations for radio access networks (RANs) and core networks in 5G/6G mobile networks. We introduce a novel performance metrics, known as FEACI, to quantitatively assess the format (F), explainability (E), accuracy (A), cost (C), and inference time (I) of the generated answer; existing general metrics are unable to capture these features. The results of our study demonstrate that open-source models can achieve comparable or even superior translation performance compared with the closed-source models requiring costly hardware setup and not accessible to all users.
Authors: Armin Iske
Abstract: We analyse the convergence of sampling algorithms for functions in reproducing kernel Hilbert spaces (RKHS). To this end, we discuss approximation properties of kernel regression under minimalistic assumptions on both the kernel and the input data. We first prove error estimates in the kernel's RKHS norm. This leads us to new results concerning uniform convergence of kernel regression on compact domains. For Lipschitz continuous and H\"older continuous kernels, we prove convergence rates.
Authors: Numan Saeed, Shahad Hardan, Muhammad Ridzuan, Nada Saadi, Karthik Nandakumar, Mohammad Yaqub
Abstract: Cancer detection and prognosis relies heavily on medical imaging, particularly CT and PET scans. Deep Neural Networks (DNNs) have shown promise in tumor segmentation by fusing information from these modalities. However, a critical bottleneck exists: the dependency on CT-PET data concurrently for training and inference, posing a challenge due to the limited availability of PET scans. Hence, there is a clear need for a flexible and efficient framework that can be trained with the widely available CT scans and can be still adapted for PET scans when they become available. In this work, we propose a parameter-efficient multi-modal adaptation (PEMMA) framework for lightweight upgrading of a transformer-based segmentation model trained only on CT scans such that it can be efficiently adapted for use with PET scans when they become available. This framework is further extended to perform prognosis task maintaining the same efficient cross-modal fine-tuning approach. The proposed approach is tested with two well-known segementation backbones, namely UNETR and Swin UNETR. Our approach offers two main advantages. Firstly, we leverage the inherent modularity of the transformer architecture and perform low-rank adaptation (LoRA) as well as decomposed low-rank adaptation (DoRA) of the attention weights to achieve parameter-efficient adaptation. Secondly, by minimizing cross-modal entanglement, PEMMA allows updates using only one modality without causing catastrophic forgetting in the other. Our method achieves comparable performance to early fusion, but with only 8% of the trainable parameters, and demonstrates a significant +28% Dice score improvement on PET scans when trained with a single modality. Furthermore, in prognosis, our method improves the concordance index by +10% when adapting a CT-pretrained model to include PET scans, and by +23% when adapting for both PET and EHR data.
Authors: Andrea Santilli, Adam Golinski, Michael Kirchhof, Federico Danieli, Arno Blaas, Miao Xiong, Luca Zappella, Sinead Williamson
Abstract: Uncertainty Quantification (UQ) in Language Models (LMs) is crucial for improving their safety and reliability. Evaluations often use performance metrics like AUROC to assess how well UQ methods (e.g., negative sequence probabilities) correlate with task correctness functions (e.g., ROUGE-L). In this paper, we show that commonly used correctness functions bias UQ evaluations by inflating the performance of certain UQ methods. We evaluate 7 correctness functions -- from lexical-based and embedding-based metrics to LLM-as-a-judge approaches -- across 4 datasets x 4 models x 6 UQ methods. Our analysis reveals that length biases in the errors of these correctness functions distort UQ assessments by interacting with length biases in UQ methods. We identify LLM-as-a-judge approaches as among the least length-biased choices and hence a potential solution to mitigate these biases.
Authors: Stefano M. Iacus, Haodong Qi, Jiyoung Han
Abstract: This paper presents a hybrid framework for literature reviews that augments traditional bibliometric methods with large language models (LLMs). By fine-tuning open-source LLMs, our approach enables scalable extraction of qualitative insights from large volumes of research content, enhancing both the breadth and depth of knowledge synthesis. To improve annotation efficiency and consistency, we introduce an error-focused validation process in which LLMs generate initial labels and human reviewers correct misclassifications. Applying this framework to over 20000 scientific articles about human migration, we demonstrate that a domain-adapted LLM can serve as a "specialist" model - capable of accurately selecting relevant studies, detecting emerging trends, and identifying critical research gaps. Notably, the LLM-assisted review reveals a growing scholarly interest in climate-induced migration. However, existing literature disproportionately centers on a narrow set of environmental hazards (e.g., floods, droughts, sea-level rise, and land degradation), while overlooking others that more directly affect human health and well-being, such as air and water pollution or infectious diseases. This imbalance highlights the need for more comprehensive research that goes beyond physical environmental changes to examine their ecological and societal consequences, particularly in shaping migration as an adaptive response. Overall, our proposed framework demonstrates the potential of fine-tuned LLMs to conduct more efficient, consistent, and insightful literature reviews across disciplines, ultimately accelerating knowledge synthesis and scientific discovery.
Authors: Gianluca Carloni
Abstract: This work aligns deep learning (DL) with human reasoning capabilities and needs to enable more efficient, interpretable, and robust image classification. We approach this from three perspectives: explainability, causality, and biological vision. Introduction and background open this work before diving into operative chapters. First, we assess neural networks' visualization techniques for medical images and validate an explainable-by-design method for breast mass classification. A comprehensive review at the intersection of XAI and causality follows, where we introduce a general scaffold to organize past and future research, laying the groundwork for our second perspective. In the causality direction, we propose novel modules that exploit feature co-occurrence in medical images, leading to more effective and explainable predictions. We further introduce CROCODILE, a general framework that integrates causal concepts, contrastive learning, feature disentanglement, and prior knowledge to enhance generalization. Lastly, we explore biological vision, examining how humans recognize objects, and propose CoCoReco, a connectivity-inspired network with context-aware attention mechanisms. Overall, our key findings include: (i) simple activation maximization lacks insight for medical imaging DL models; (ii) prototypical-part learning is effective and radiologically aligned; (iii) XAI and causal ML are deeply connected; (iv) weak causal signals can be leveraged without a priori information to improve performance and interpretability; (v) our framework generalizes across medical domains and out-of-distribution data; (vi) incorporating biological circuit motifs improves human-aligned recognition. This work contributes toward human-aligned DL and highlights pathways to bridge the gap between research and clinical adoption, with implications for improved trust, diagnostic accuracy, and safe deployment.
Authors: Paul K. Mandal, Cole Leo, Connor Hurley
Abstract: Open-source intelligence provides a stream of unstructured textual data that can inform assessments of territorial control. We present CONTACT, a framework for territorial control prediction using large language models (LLMs) and minimal supervision. We evaluate two approaches: SetFit, an embedding-based few-shot classifier, and a prompt tuning method applied to BLOOMZ-560m, a multilingual generative LLM. Our model is trained on a small hand-labeled dataset of news articles covering ISIS activity in Syria and Iraq, using prompt-conditioned extraction of control-relevant signals such as military operations, casualties, and location references. We show that the BLOOMZ-based model outperforms the SetFit baseline, and that prompt-based supervision improves generalization in low-resource settings. CONTACT demonstrates that LLMs fine-tuned using few-shot methods can reduce annotation burdens and support structured inference from open-ended OSINT streams. Our code is available at https://github.com/PaulKMandal/CONTACT/.
Authors: Davide Ghiani, Jefferson David Rodriguez Chivata, Stefano Lilliu, Simone Maurizio La Cava, Marco Micheletto, Giulia Orr\`u, Federico Lama, Gian Luca Marcialis
Abstract: Modern identity verification systems increasingly rely on facial images embedded in biometric documents such as electronic passports. To ensure global interoperability and security, these images must comply with strict standards defined by the International Civil Aviation Organization (ICAO), which specify acquisition, quality, and format requirements. However, once issued, these images may undergo unintentional degradations (e.g., compression, resizing) or malicious manipulations (e.g., morphing) and deceive facial recognition systems. In this study, we explore fragile watermarking, based on deep steganographic embedding as a proactive mechanism to certify the authenticity of ICAO-compliant facial images. By embedding a hidden image within the official photo at the time of issuance, we establish an integrity marker that becomes sensitive to any post-issuance modification. We assess how a range of image manipulations affects the recovered hidden image and show that degradation artifacts can serve as robust forensic cues. Furthermore, we propose a classification framework that analyzes the revealed content to detect and categorize the type of manipulation applied. Our experiments demonstrate high detection accuracy, including cross-method scenarios with multiple deep steganography-based models. These findings support the viability of fragile watermarking via steganographic embedding as a valuable tool for biometric document integrity verification.
Authors: Robert Busa-Fekete, Umar Syed
Abstract: We present new algorithms for estimating and testing \emph{collision probability}, a fundamental measure of the spread of a discrete distribution that is widely used in many scientific fields. We describe an algorithm that satisfies $(\alpha, \beta)$-local differential privacy and estimates collision probability with error at most $\epsilon$ using $\tilde{O}\left(\frac{\log(1/\beta)}{\alpha^2 \epsilon^2}\right)$ samples for $\alpha \le 1$, which improves over previous work by a factor of $\frac{1}{\alpha^2}$. We also present a sequential testing algorithm for collision probability, which can distinguish between collision probability values that are separated by $\epsilon$ using $\tilde{O}(\frac{1}{\epsilon^2})$ samples, even when $\epsilon$ is unknown. Our algorithms have nearly the optimal sample complexity, and in experiments we show that they require significantly fewer samples than previous methods.
Authors: Feijiang Han, Jiaming Zhang, Chuyi Deng, Jianheng Tang, Yunhuai Liu
Abstract: WebShell attacks, in which malicious scripts are injected into web servers, are a major cybersecurity threat. Traditional machine learning and deep learning methods are hampered by issues such as the need for extensive training data, catastrophic forgetting, and poor generalization. Recently, Large Language Models (LLMs) have gained attention for code-related tasks, but their potential in WebShell detection remains underexplored. In this paper, we make two major contributions: (1) a comprehensive evaluation of seven LLMs, including GPT-4, LLaMA 3.1 70B, and Qwen 2.5 variants, benchmarked against traditional sequence- and graph-based methods using a dataset of 26.59K PHP scripts, and (2) the Behavioral Function-Aware Detection (BFAD) framework, designed to address the specific challenges of applying LLMs to this domain. Our framework integrates three components: a Critical Function Filter that isolates malicious PHP function calls, a Context-Aware Code Extraction strategy that captures the most behaviorally indicative code segments, and Weighted Behavioral Function Profiling (WBFP) that enhances in-context learning by prioritizing the most relevant demonstrations based on discriminative function-level profiles. Our results show that larger LLMs achieve near-perfect precision but lower recall, while smaller models exhibit the opposite trade-off. However, all models lag behind previous State-Of-The-Art (SOTA) methods. With BFAD, the performance of all LLMs improved, with an average F1 score increase of 13.82%. Larger models such as GPT-4, LLaMA 3.1 70B, and Qwen 2.5 14B outperform SOTA methods, while smaller models such as Qwen 2.5 3B achieve performance competitive with traditional methods. This work is the first to explore the feasibility and limitations of LLMs for WebShell detection, and provides solutions to address the challenges in this task.
Authors: Junjie Yang, Junhao Song, Xudong Han, Ziqian Bi, Tianyang Wang, Chia Xin Liang, Xinyuan Song, Yichao Zhang, Qian Niu, Benji Peng, Keyu Chen, Ming Liu
Abstract: Knowledge distillation (KD) is a technique for transferring knowledge from complex teacher models to simpler student models, significantly enhancing model efficiency and accuracy. It has demonstrated substantial advancements in various applications including image classification, object detection, language modeling, text classification, and sentiment analysis. Recent innovations in KD methods, such as attention-based approaches, block-wise logit distillation, and decoupling distillation, have notably improved student model performance. These techniques focus on stimulus complexity, attention mechanisms, and global information capture to optimize knowledge transfer. In addition, KD has proven effective in compressing large language models while preserving accuracy, reducing computational overhead, and improving inference speed. This survey synthesizes the latest literature, highlighting key findings, contributions, and future directions in knowledge distillation to provide insights for researchers and practitioners on its evolving role in artificial intelligence and machine learning.
Authors: Andi Nika, Sepehr Elahi, \c{C}a\u{g}{\i}n Ararat, Cem Tekin
Abstract: We consider the problem of optimizing a vector-valued objective function $\boldsymbol{f}$ sampled from a Gaussian Process (GP) whose index set is a well-behaved, compact metric space $({\cal X},d)$ of designs. We assume that $\boldsymbol{f}$ is not known beforehand and that evaluating $\boldsymbol{f}$ at design $x$ results in a noisy observation of $\boldsymbol{f}(x)$. Since identifying the Pareto optimal designs via exhaustive search is infeasible when the cardinality of ${\cal X}$ is large, we propose an algorithm, called Adaptive $\boldsymbol{\epsilon}$-PAL, that exploits the smoothness of the GP-sampled function and the structure of $({\cal X},d)$ to learn fast. In essence, Adaptive $\boldsymbol{\epsilon}$-PAL employs a tree-based adaptive discretization technique to identify an $\boldsymbol{\epsilon}$-accurate Pareto set of designs in as few evaluations as possible. We provide both information-type and metric dimension-type bounds on the sample complexity of $\boldsymbol{\epsilon}$-accurate Pareto set identification. We also experimentally show that our algorithm outperforms other Pareto set identification methods.
Authors: Han-Dong Lim, Donghwan Lee
Abstract: Off-policy learning ability is an important feature of reinforcement learning (RL) for practical applications. However, even one of the most elementary RL algorithms, temporal-difference (TD) learning, is known to suffer form divergence issue when the off-policy scheme is used together with linear function approximation. To overcome the divergent behavior, several off-policy TD-learning algorithms, including gradient-TD learning (GTD), and TD-learning with correction (TDC), have been developed until now. In this work, we provide a unified view of such algorithms from a purely control-theoretic perspective, and propose a new convergent algorithm. Our method relies on the backstepping technique, which is widely used in nonlinear control theory. Finally, convergence of the proposed algorithm is experimentally verified in environments where the standard TD-learning is known to be unstable.
Authors: Ahsan Adeel, Junaid Muzaffar, Fahad Zia, Khubaib Ahmed, Mohsin Raza, Eamin Chaudary, Talha Bin Riaz, Ahmed Saeed
Abstract: Going beyond 'dendritic democracy', we introduce a 'democracy of local processors', termed Cooperator. Here we compare their capabilities when used in permutation invariant neural networks for reinforcement learning (RL), with machine learning algorithms based on Transformers, such as ChatGPT. Transformers are based on the long standing conception of integrate-and-fire 'point' neurons, whereas Cooperator is inspired by recent neurobiological breakthroughs suggesting that the cellular foundations of mental life depend on context-sensitive pyramidal neurons in the neocortex which have two functionally distinct points. Weshow that when used for RL, an algorithm based on Cooperator learns far quicker than that based on Transformer, even while having the same number of parameters.
Authors: Mircea Petrache, Shubhendu Trivedi
Abstract: The explicit incorporation of task-specific inductive biases through symmetry has emerged as a general design precept in the development of high-performance machine learning models. For example, group equivariant neural networks have demonstrated impressive performance across various domains and applications such as protein and drug design. A prevalent intuition about such models is that the integration of relevant symmetry results in enhanced generalization. Moreover, it is posited that when the data and/or the model may only exhibit $\textit{approximate}$ or $\textit{partial}$ symmetry, the optimal or best-performing model is one where the model symmetry aligns with the data symmetry. In this paper, we conduct a formal unified investigation of these intuitions. To begin, we present general quantitative bounds that demonstrate how models capturing task-specific symmetries lead to improved generalization. In fact, our results do not require the transformations to be finite or even form a group and can work with partial or approximate equivariance. Utilizing this quantification, we examine the more general question of model mis-specification i.e. when the model symmetries don't align with the data symmetries. We establish, for a given symmetry group, a quantitative comparison between the approximate/partial equivariance of the model and that of the data distribution, precisely connecting model equivariance error and data equivariance error. Our result delineates conditions under which the model equivariance error is optimal, thereby yielding the best-performing model for the given task and data. Our results are the most general results of their type in the literature.
Authors: Alexandru Dumitrescu, Dani Korpela, Markus Heinonen, Yogesh Verma, Valerii Iakovlev, Vikas Garg, Harri L\"ahdesm\"aki
Abstract: Obtaining the desired effect of drugs is highly dependent on their molecular geometries. Thus, the current prevailing paradigm focuses on 3D point-cloud atom representations, utilizing graph neural network (GNN) parametrizations, with rotational symmetries baked in via E(3) invariant layers. We prove that such models must necessarily disregard chirality, a geometric property of the molecules that cannot be superimposed on their mirror image by rotation and translation. Chirality plays a key role in determining drug safety and potency. To address this glaring issue, we introduce a novel field-based representation, proposing reference rotations that replace rotational symmetry constraints. The proposed model captures all molecular geometries including chirality, while still achieving highly competitive performance with E(3)-based methods across standard benchmarking metrics.
Authors: Haotian Chen, Anna Kuzina, Babak Esmaeili, Jakub M Tomczak
Abstract: Current state-of-the-art optimizers are adaptive gradient-based optimization methods such as Adam. Recently, there has been an increasing interest in formulating gradient-based optimizers in a probabilistic framework for better modeling the uncertainty of the gradients. Here, we propose to combine both approaches, resulting in the Variational Stochastic Gradient Descent (VSGD) optimizer. We model gradient updates as a probabilistic model and utilize stochastic variational inference (SVI) to derive an efficient and effective update rule. Further, we show how our VSGD method relates to other adaptive gradient-based optimizers like Adam. Lastly, we carry out experiments on two image classification datasets and four deep neural network architectures, where we show that VSGD outperforms Adam and SGD.
Authors: Haoxian Chen, Hanyang Zhao, Henry Lam, David Yao, Wenpin Tang
Abstract: Direct Preference Optimization (DPO) has recently emerged as a popular approach to improve reinforcement learning with human feedback (RLHF), leading to better techniques to fine-tune large language models (LLM). A weakness of DPO, however, lies in its lack of capability to characterize the diversity of human preferences. Inspired by Mallows' theory of preference ranking, we develop in this paper a new approach, the MallowsPO. A distinct feature of this approach is a dispersion index, which reflects the dispersion of human preference to prompts. We show that existing DPO models can be reduced to special cases of this dispersion index, thus unified with MallowsPO. More importantly, we demonstrate (empirically) how to use this dispersion index to enhance the performance of DPO in a broad array of benchmark tasks, from synthetic bandit selection to controllable generations and dialogues, while maintaining great generalization capabilities. MallowsPO is also compatible with other SOTA offline preference optimization methods, boosting nearly 2\% extra LC win rate when used as a plugin for fine-tuning Llama3-Instruct.
Authors: Patrick Emami, Zhaonan Li, Saumya Sinha, Truc Nguyen
Abstract: Surrogate models are used to predict the behavior of complex energy systems that are too expensive to simulate with traditional numerical methods. Our work introduces the use of language descriptions, which we call ``system captions'' or SysCaps, to interface with such surrogates. We argue that interacting with surrogates through text, particularly natural language, makes these models more accessible for both experts and non-experts. We introduce a lightweight multimodal text and timeseries regression model and a training pipeline that uses large language models (LLMs) to synthesize high-quality captions from simulation metadata. Our experiments on two real-world simulators of buildings and wind farms show that our SysCaps-augmented surrogates have better accuracy on held-out systems than traditional methods while enjoying new generalization abilities, such as handling semantically related descriptions of the same test system. Additional experiments also highlight the potential of SysCaps to unlock language-driven design space exploration and to regularize training through prompt augmentation.
Authors: Daniel P. Jeong, Zachary C. Lipton, Pradeep Ravikumar
Abstract: In this paper, we demonstrate a surprising capability of large language models (LLMs): given only input feature names and a description of a prediction task, they are capable of selecting the most predictive features, with performance rivaling the standard tools of data science. Remarkably, these models exhibit this capacity across various query mechanisms. For example, we zero-shot prompt an LLM to output a numerical importance score for a feature (e.g., "blood pressure") in predicting an outcome of interest (e.g., "heart failure"), with no additional context. In particular, we find that the latest models, such as GPT-4, can consistently identify the most predictive features regardless of the query mechanism and across various prompting strategies. We illustrate these findings through extensive experiments on real-world data, where we show that LLM-based feature selection consistently achieves strong performance competitive with data-driven methods such as the LASSO, despite never having looked at the downstream training data. Our findings suggest that LLMs may be useful not only for selecting the best features for training but also for deciding which features to collect in the first place. This could benefit practitioners in domains like healthcare and the social sciences, where collecting high-quality data comes at a high cost.
Authors: Junkang Wu, Yuexiang Xie, Zhengyi Yang, Jiancan Wu, Jiawei Chen, Jinyang Gao, Bolin Ding, Xiang Wang, Xiangnan He
Abstract: This study addresses the challenge of noise in training datasets for Direct Preference Optimization (DPO), a method for aligning Large Language Models (LLMs) with human preferences. We categorize noise into pointwise noise, which includes low-quality data points, and pairwise noise, which encompasses erroneous data pair associations that affect preference rankings. Utilizing Distributionally Robust Optimization (DRO), we enhance DPO's resilience to these types of noise. Our theoretical insights reveal that DPO inherently embeds DRO principles, conferring robustness to pointwise noise, with the regularization coefficient $\beta$ playing a critical role in its noise resistance. Extending this framework, we introduce Distributionally Robustifying DPO (Dr. DPO), which integrates pairwise robustness by optimizing against worst-case pairwise scenarios. The novel hyperparameter $\beta'$ in Dr. DPO allows for fine-tuned control over data pair reliability, providing a strategic balance between exploration and exploitation in noisy training environments. Empirical evaluations demonstrate that Dr. DPO substantially improves the quality of generated text and response accuracy in preference datasets, showcasing enhanced performance in both noisy and noise-free settings. The code is available at https://github.com/junkangwu/Dr_DPO.
Authors: Callen MacPhee, Yiming 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 face challenges such as high memory requirements and computational costs with limited interpretability. This paper introduces an approach where master equations of physics are converted into multilayered networks that are trained via backpropagation. The resulting general-purpose model effectively encodes data in the properties of the underlying physical system. In contrast to existing methods wherein a trained neural network is used as a computationally efficient alternative for solving physical equations, our approach directly treats physics equations as trainable models. We demonstrate this physical embedding concept with the Nonlinear Schr\"odinger Equation (NLSE), which acts as trainable architecture for learning complex patterns including nonlinear mappings and memory effects from data. The network embeds data representation in orders of magnitude fewer parameters than conventional neural networks when tested on time series data. Notably, the trained "Nonlinear Schr\"odinger Network" is interpretable, with all parameters having physical meanings. This interpretability offers insight into the underlying dynamics of the system that produced the data. The proposed method of replacing traditional DNN feature learning architectures with physical equations is also extended to the Gross-Pitaevskii Equation, demonstrating the broad applicability of the framework to other master equations of physics. Among our results, an ablation study quantifies the relative importance of physical terms such as dispersion, nonlinearity, and potential energy for classification accuracy. We also outline the limitations of this approach as it relates to generalizability.
Authors: Yupeng Chen, Senmiao Wang, Yushun Zhang, Zhihang Lin, Haozhe Zhang, Weijian Sun, Tian Ding, Ruoyu Sun
Abstract: Large language models (LLMs) have demonstrated remarkable capabilities across a wide range of tasks. Typically, LLMs are first pre-trained on large corpora and subsequently fine-tuned on task-specific datasets. However, during fine-tuning, LLMs may forget some knowledge acquired in the pre-training stage, leading to a decline in general capabilities. Existing approaches to mitigate forgetting often rely on access to pre-training data, which may be unavailable in many real-world scenarios--such as fine-tuning checkpoint-only open-source LLMs. To address this challenge, we propose a new fine-tuning algorithm termed Momentum-Filtered Optimizer (MoFO). MoFO is an extension of greedy block coordinate descent (BCD) methods: in each iteration, MoFO only updates the model parameters with the largest momentum magnitudes, while keeping all other parameters fixed. MoFO achieves similar fine-tuning performance to the default fine-tuning algorithm while effectively mitigating knowledge forgetting. We validate MoFO through rigorous convergence analysis and extensive experiments, demonstrating its effectiveness in mitigating forgetting without pre-training data.
Authors: Andrew Kiruluta, Andreas Lemos
Abstract: This paper presents a novel hybrid approach that integrates linear programming (LP) within the loss function of an unsupervised machine learning model. By leveraging the strengths of both optimization techniques and machine learning, this method introduces a robust framework for solving complex optimization problems where traditional methods may fall short. The proposed approach encapsulates the constraints and objectives of a linear programming problem directly into the loss function, guiding the learning process to adhere to these constraints while optimizing the desired outcomes. This technique not only preserves the interpretability of linear programming but also benefits from the flexibility and adaptability of machine learning, making it particularly well-suited for unsupervised or semi-supervised learning scenarios.
Authors: Sihang Li, Jin Huang, Jiaxi Zhuang, Yaorui Shi, Xiaochen Cai, Mingjun Xu, Xiang Wang, Linfeng Zhang, Guolin Ke, Hengxing Cai
Abstract: Scientific literature understanding is crucial for extracting targeted information and garnering insights, thereby significantly advancing scientific discovery. Despite the remarkable success of Large Language Models (LLMs), they face challenges in scientific literature understanding, primarily due to (1) a lack of scientific knowledge and (2) unfamiliarity with specialized scientific tasks. To develop an LLM specialized in scientific literature understanding, we propose a hybrid strategy that integrates continual pre-training (CPT) and supervised fine-tuning (SFT), to simultaneously infuse scientific domain knowledge and enhance instruction-following capabilities for domain-specific tasks.cIn this process, we identify two key challenges: (1) constructing high-quality CPT corpora, and (2) generating diverse SFT instructions. We address these challenges through a meticulous pipeline, including PDF text extraction, parsing content error correction, quality filtering, and synthetic instruction creation. Applying this strategy, we present a suite of LLMs: SciLitLLM, specialized in scientific literature understanding. These models demonstrate promising performance on scientific literature understanding benchmarks. Our contributions are threefold: (1) We present an effective framework that integrates CPT and SFT to adapt LLMs to scientific literature understanding, which can also be easily adapted to other domains. (2) We propose an LLM-based synthesis method to generate diverse and high-quality scientific instructions, resulting in a new instruction set -- SciLitIns -- for supervised fine-tuning in less-represented scientific domains. (3) SciLitLLM achieves promising performance improvements on scientific literature understanding benchmarks.
Authors: Jo\~ao Bravo
Abstract: Despite the rise to dominance of deep learning in unstructured data domains, tree-based methods such as Random Forests (RF) and Gradient Boosted Decision Trees (GBDT) are still the workhorses for handling discriminative tasks on tabular data. We explore generative extensions of these popular algorithms with a focus on explicitly modeling the data density (up to a normalization constant), thus enabling other applications besides sampling. As our main contribution we propose an energy-based generative boosting algorithm that is analogous to the second-order boosting implemented in popular libraries like XGBoost. We show that, despite producing a generative model capable of handling inference tasks over any input variable, our proposed algorithm can achieve similar discriminative performance to GBDT on a number of real world tabular datasets, outperforming alternative generative approaches. At the same time, we show that it is also competitive with neural-network-based models for sampling. Code is available at https://github.com/ajoo/nrgboost.
Authors: A. Quadir, M. Sajid, M. Tanveer
Abstract: On Efficient and Scalable Computation of the Nonparametric Maximum Likelihood Estimator in Mixture ModelsTwin support vector machine (TSVM) is an emerging machine learning model with versatile applicability in classification and regression endeavors. Nevertheless, TSVM confronts noteworthy challenges: $(i)$ the imperative demand for matrix inversions presents formidable obstacles to its efficiency and applicability on large-scale datasets; $(ii)$ the omission of the structural risk minimization (SRM) principle in its primal formulation heightens the vulnerability to overfitting risks; and $(iii)$ the TSVM exhibits a high susceptibility to noise and outliers, and also demonstrates instability when subjected to resampling. In view of the aforementioned challenges, we propose the granular ball twin support vector machine (GBTSVM). GBTSVM takes granular balls, rather than individual data points, as inputs to construct a classifier. These granular balls, characterized by their coarser granularity, exhibit robustness to resampling and reduced susceptibility to the impact of noise and outliers. We further propose a novel large-scale granular ball twin support vector machine (LS-GBTSVM). LS-GBTSVM's optimization formulation ensures two critical facets: $(i)$ it eliminates the need for matrix inversions, streamlining the LS-GBTSVM's computational efficiency, and $(ii)$ it incorporates the SRM principle through the incorporation of regularization terms, effectively addressing the issue of overfitting. The proposed LS-GBTSVM exemplifies efficiency, scalability for large datasets, and robustness against noise and outliers. We conduct a comprehensive evaluation of the GBTSVM and LS-GBTSVM models on benchmark datasets from UCI, KEEL, and NDC datasets. Our experimental findings and statistical analyses affirm the superior generalization prowess of the proposed GBTSVM and LS-GBTSVM models.
Authors: Maksym Andriushchenko, Alexandra Souly, Mateusz Dziemian, Derek Duenas, Maxwell Lin, Justin Wang, Dan Hendrycks, Andy Zou, Zico Kolter, Matt Fredrikson, Eric Winsor, Jerome Wynne, Yarin Gal, Xander Davies
Abstract: The robustness of LLMs to jailbreak attacks, where users design prompts to circumvent safety measures and misuse model capabilities, has been studied primarily for LLMs acting as simple chatbots. Meanwhile, LLM agents -- which use external tools and can execute multi-stage tasks -- may pose a greater risk if misused, but their robustness remains underexplored. To facilitate research on LLM agent misuse, we propose a new benchmark called AgentHarm. The benchmark includes a diverse set of 110 explicitly malicious agent tasks (440 with augmentations), covering 11 harm categories including fraud, cybercrime, and harassment. In addition to measuring whether models refuse harmful agentic requests, scoring well on AgentHarm requires jailbroken agents to maintain their capabilities following an attack to complete a multi-step task. We evaluate a range of leading LLMs, and find (1) leading LLMs are surprisingly compliant with malicious agent requests without jailbreaking, (2) simple universal jailbreak templates can be adapted to effectively jailbreak agents, and (3) these jailbreaks enable coherent and malicious multi-step agent behavior and retain model capabilities. To enable simple and reliable evaluation of attacks and defenses for LLM-based agents, we publicly release AgentHarm at https://huggingface.co/datasets/ai-safety-institute/AgentHarm.
URLs: https://huggingface.co/datasets/ai-safety-institute/AgentHarm.
Authors: Zifei Xu, Sayeh Sharify, Wanzin Yazar, Tristan Webb, Xin Wang
Abstract: Large language models of high parameter counts are computationally expensive, yet can be made much more efficient by compressing their weights to very low numerical precision. This can be achieved either through post-training quantization by minimizing local, layer-wise quantization errors, or through quantization-aware fine-tuning by minimizing the global loss function. In this study, we discovered that, under the same data constraint, the former approach nearly always fared worse than the latter, a phenomenon particularly prominent when the numerical precision is very low. We further showed that this difficulty of post-training quantization arose from stark misalignment between optimization of the local and global objective functions. Our findings explains limited utility in minimization of local quantization error and the importance of direct quantization-aware fine-tuning, in the regime of large models at very low precision.
Authors: M. Tanveer, A. Quadir
Abstract: One of the major difficulties in machine learning methods is categorizing datasets that are imbalanced. This problem may lead to biased models, where the training process is dominated by the majority class, resulting in inadequate representation of the minority class. Universum twin support vector machine (UTSVM) produces a biased model towards the majority class, as a result, its performance on the minority class is often poor as it might be mistakenly classified as noise. Moreover, UTSVM is not proficient in handling datasets that contain outliers and noises. Inspired by the concept of incorporating prior information about the data and employing an intuitionistic fuzzy membership scheme, we propose intuitionistic fuzzy UTSVM for imbalanced data (IFUTSVM-ID) by enhancing overall robustness. We use an intuitionistic fuzzy membership scheme to mitigate the impact of noise and outliers. Moreover, to tackle the problem of imbalanced class distribution, data oversampling and undersampling methods are utilized. Prior knowledge about the data is provided by universum data. This leads to better generalization performance. UTSVM is susceptible to overfitting risks due to the omission of the structural risk minimization (SRM) principle in their primal formulations. However, the proposed IFUTSVM-ID model incorporates the SRM principle through the incorporation of regularization terms, effectively addressing the issue of overfitting. We conduct a comprehensive evaluation of the proposed IFUTSVM-ID model on benchmark datasets from KEEL and compare it with existing baseline models. Furthermore, to assess the effectiveness of the proposed IFUTSVM-ID model in diagnosing Alzheimer's disease (AD), we applied them to the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. Experimental results showcase the superiority of the proposed IFUTSVM-ID models compared to the baseline models.
Authors: Hanshi Sun, Li-Wen Chang, Wenlei Bao, Size Zheng, Ningxin Zheng, Xin Liu, Harry Dong, Yuejie Chi, Beidi Chen
Abstract: With the widespread deployment of long-context large language models (LLMs), there has been a growing demand for efficient support of high-throughput inference. However, as the key-value (KV) cache expands with the sequence length, the increasing memory footprint and the need to access it for each token generation both result in low throughput when serving long-context LLMs. While various dynamic sparse attention methods have been proposed to speed up inference while maintaining generation quality, they either fail to sufficiently reduce GPU memory consumption or introduce significant decoding latency by offloading the KV cache to the CPU. We present ShadowKV, a high-throughput long-context LLM inference system that stores the low-rank key cache and offloads the value cache to reduce the memory footprint for larger batch sizes and longer sequences. To minimize decoding latency, ShadowKV employs an accurate KV selection strategy that reconstructs minimal sparse KV pairs on-the-fly. By evaluating ShadowKV on a broad range of benchmarks, including RULER, LongBench, and Needle In A Haystack, and models like Llama-3.1-8B, Llama-3-8B-1M, GLM-4-9B-1M, Yi-9B-200K, Phi-3-Mini-128K, and Qwen2-7B-128K, we demonstrate that it can support up to 6$\times$ larger batch sizes and boost throughput by up to 3.04$\times$ on an A100 GPU without sacrificing accuracy, even surpassing the performance achievable with infinite batch size under the assumption of infinite GPU memory. The code is available at https://github.com/bytedance/ShadowKV.
Authors: Bowen Liu, Haoyang Li, Shuning Wang, Shuo Nie, Shanghang Zhang
Abstract: Out-of-distribution (OOD) generalization in Graph Neural Networks (GNNs) has gained significant attention due to its critical importance in graph-based predictions in real-world scenarios. Existing methods primarily focus on extracting a single causal subgraph from the input graph to achieve generalizable predictions. However, relying on a single subgraph can lead to susceptibility to spurious correlations and is insufficient for learning invariant patterns behind graph data. Moreover, in many real-world applications, such as molecular property prediction, multiple critical subgraphs may influence the target label property. To address these challenges, we propose a novel framework, SubGraph Aggregation (SuGAr), designed to learn a diverse set of subgraphs that are crucial for OOD generalization on graphs. Specifically, SuGAr employs a tailored subgraph sampler and diversity regularizer to extract a diverse set of invariant subgraphs. These invariant subgraphs are then aggregated by averaging their representations, which enriches the subgraph signals and enhances coverage of the underlying causal structures, thereby improving OOD generalization. Extensive experiments on both synthetic and real-world datasets demonstrate that \ours outperforms state-of-the-art methods, achieving up to a 24% improvement in OOD generalization on graphs. To the best of our knowledge, this is the first work to study graph OOD generalization by learning multiple invariant subgraphs. code: https://github.com/Nanolbw/SuGAr
Authors: Hui Zhang, Fu Liu
Abstract: Among adversarial attacks against sequential recommender systems, model extraction attacks represent a method to attack sequential recommendation models without prior knowledge. Existing research has primarily concentrated on the adversary's execution of black-box attacks through data-free model extraction. However, a significant gap remains in the literature concerning the development of surrogate models by adversaries with access to few-shot raw data (10\% even less). That is, the challenge of how to construct a surrogate model with high functional similarity within the context of few-shot data scenarios remains an issue that requires resolution.This study addresses this gap by introducing a novel few-shot model extraction framework against sequential recommenders, which is designed to construct a superior surrogate model with the utilization of few-shot data. The proposed few-shot model extraction framework is comprised of two components: an autoregressive augmentation generation strategy and a bidirectional repair loss-facilitated model distillation procedure. Specifically, to generate synthetic data that closely approximate the distribution of raw data, autoregressive augmentation generation strategy integrates a probabilistic interaction sampler to extract inherent dependencies and a synthesis determinant signal module to characterize user behavioral patterns. Subsequently, bidirectional repair loss, which target the discrepancies between the recommendation lists, is designed as auxiliary loss to rectify erroneous predictions from surrogate models, transferring knowledge from the victim model to the surrogate model effectively. Experiments on three datasets show that the proposed few-shot model extraction framework yields superior surrogate models.
Authors: Yiqun Zhang, Mingjie Zhao, Hong Jia, Yang Lu, Mengke Li, Yiu-ming Cheung
Abstract: Categorical data composed of qualitative valued attributes are ubiquitous in machine learning tasks. Due to the lack of well-defined metric space, categorical data distributions are difficult to be intuitively understood. Clustering is a popular data analysis technique suitable for data distribution understanding. However, the success of clustering often relies on reasonable distance metrics, which happens to be what categorical data naturally lack. This paper therefore introduces a new finding that the order relation among attribute values is the decisive factor in clustering accuracy, and is also the key to understanding categorical data clusters, because the essence of clustering is to order the clusters in terms of their admission to samples. To obtain the orders, we propose a new learning paradigm that allows joint learning of clusters and the orders. It alternatively partitions the data into clusters based on the distance metric built upon the orders and estimates the most likely orders according to the clusters. The algorithm achieves superior clustering accuracy with a convergence guarantee, and the learned orders facilitate the understanding of the non-intuitive cluster distribution of categorical data. Extensive experiments with ablation studies, statistical evidence, and case studies have validated the new insight into the importance of value order and the method proposition. The source code is temporarily opened in https://anonymous.4open.science/r/OCL-demo.
Authors: Amit Kumar Bhuyan, Hrishikesh Dutta, Subir Biswas
Abstract: This paper introduces an Unmanned Aerial Vehicle - enabled content management architecture that is suitable for critical content access in communities of users that are communication-isolated during diverse types of disaster scenarios. The proposed architecture leverages a hybrid network of stationary anchor UAVs and mobile Micro-UAVs for ubiquitous content dissemination. The anchor UAVs are equipped with both vertical and lateral communication links, and they serve local users, while the mobile micro-ferrying UAVs extend coverage across communities with increased mobility. The focus is on developing a content dissemination system that dynamically learns optimal caching policies to maximize content availability. The core innovation is an adaptive content dissemination framework based on distributed Federated Multi-Armed Bandit learning. The goal is to optimize UAV content caching decisions based on geo-temporal content popularity and user demand variations. A Selective Caching Algorithm is also introduced to reduce redundant content replication by incorporating inter-UAV information sharing. This method strategically preserves the uniqueness in user preferences while amalgamating the intelligence across a distributed learning system. This approach improves the learning algorithm's ability to adapt to diverse user preferences. Functional verification and performance evaluation confirm the proposed architecture's utility across different network sizes, UAV swarms, and content popularity patterns.
Authors: Jielong Yang, Daoyuan Huang
Abstract: Reinforcement learning plays a crucial role in vehicle control by guiding agents to learn optimal control strategies through designing or learning appropriate reward signals. However, in vehicle control applications, rewards typically need to be manually designed while considering multiple implicit factors, which easily introduces human biases. Although imitation learning methods does not rely on explicit reward signals, they necessitate high-quality expert actions, which are often challenging to acquire. To address these issues, we propose a reward-free reinforcement learning framework (RFRLF). This framework directly learns the target states to optimize agent behavior through a target state prediction network (TSPN) and a reward-free state-guided policy network (RFSGPN), avoiding the dependence on manually designed reward signals. Specifically, the policy network is learned via minimizing the differences between the predicted state and the expert state. Experimental results demonstrate the effectiveness of the proposed RFRLF in controlling vehicle driving, showing its advantages in improving learning efficiency and adapting to reward-free environments.
Authors: Christoph Schuhmann, Gollam Rabby, Ameya Prabhu, Tawsif Ahmed, Andreas Hochlehnert, Huu Nguyen, Nick Akinci, Ludwig Schmidt, Robert Kaczmarczyk, S\"oren Auer, Jenia Jitsev, Matthias Bethge
Abstract: Paywalls, licenses and copyright rules often restrict the broad dissemination and reuse of scientific knowledge. We take the position that it is both legally and technically feasible to extract the scientific knowledge in scholarly texts. Current methods, like text embeddings, fail to reliably preserve factual content, and simple paraphrasing may not be legally sound. We propose a new idea for the community to adopt: convert scholarly documents into knowledge preserving, but style agnostic representations we term Knowledge Units using LLMs. These units use structured data capturing entities, attributes and relationships without stylistic content. We provide evidence that Knowledge Units (1) form a legally defensible framework for sharing knowledge from copyrighted research texts, based on legal analyses of German copyright law and U.S. Fair Use doctrine, and (2) preserve most (~95\%) factual knowledge from original text, measured by MCQ performance on facts from the original copyrighted text across four research domains. Freeing scientific knowledge from copyright promises transformative benefits for scientific research and education by allowing language models to reuse important facts from copyrighted text. To support this, we share open-source tools for converting research documents into Knowledge Units. Overall, our work posits the feasibility of democratizing access to scientific knowledge while respecting copyright.
Authors: Yihan Zhou, Eric Price, Trung Nguyen
Abstract: We address the problem of active logistic regression in the realizable setting. It is well known that active learning can require exponentially fewer label queries compared to passive learning, in some cases using $\log \frac{1}{\eps}$ rather than $\poly(1/\eps)$ labels to get error $\eps$ larger than the optimum. We present the first algorithm that is polynomially competitive with the optimal algorithm on every input instance, up to factors polylogarithmic in the error and domain size. In particular, if any algorithm achieves label complexity polylogarithmic in $\eps$, so does ours. Our algorithm is based on efficient sampling and can be extended to learn more general class of functions. We further support our theoretical results with experiments demonstrating performance gains for logistic regression compared to existing active learning algorithms.
Authors: Siyuan Mu, Sen Lin
Abstract: Artificial intelligence (AI) has achieved astonishing successes in many domains, especially with the recent breakthroughs in the development of foundational large models. These large models, leveraging their extensive training data, provide versatile solutions for a wide range of downstream tasks. However, as modern datasets become increasingly diverse and complex, the development of large AI models faces two major challenges: (1) the enormous consumption of computational resources and deployment difficulties, and (2) the difficulty in fitting heterogeneous and complex data, which limits the usability of the models. Mixture of Experts (MoE) models has recently attracted much attention in addressing these challenges, by dynamically selecting and activating the most relevant sub-models to process input data. It has been shown that MoEs can significantly improve model performance and efficiency with fewer resources, particularly excelling in handling large-scale, multimodal data. Given the tremendous potential MoE has demonstrated across various domains, it is urgent to provide a comprehensive summary of recent advancements of MoEs in many important fields. Existing surveys on MoE have their limitations, e.g., being outdated or lacking discussion on certain key areas, and we aim to address these gaps. In this paper, we first introduce the basic design of MoE, including gating functions, expert networks, routing mechanisms, training strategies, and system design. We then explore the algorithm design of MoE in important machine learning paradigms such as continual learning, meta-learning, multi-task learning, and reinforcement learning. Additionally, we summarize theoretical studies aimed at understanding MoE and review its applications in computer vision and natural language processing. Finally, we discuss promising future research directions.
Authors: Masoud Hashemi, Oluwanifemi Bamgbose, Sathwik Tejaswi Madhusudhan, Jishnu Sethumadhavan Nair, Aman Tiwari, Vikas Yadav
Abstract: Test-time scaling has significantly improved large language model performance, enabling deeper reasoning to solve complex problems. However, this increased reasoning capability also leads to excessive token generation and unnecessary problem-solving attempts. We introduce Don\'t Reason Bench (DNR Bench), a new benchmark designed to evaluate LLMs ability to robustly understand the tricky reasoning triggers and avoiding unnecessary generation. DNR Bench consists of 150 adversarially designed prompts that are easy for humans to understand and respond to, but surprisingly not for many of the recent prominent LLMs. DNR Bench tests models abilities across different capabilities, such as instruction adherence, hallucination avoidance, redundancy filtering, and unanswerable question recognition. We evaluate reasoning LLMs (RLMs), including DeepSeek-R1, OpenAI O3-mini, Claude-3.7-sonnet and compare them against a powerful non-reasoning model, e.g., GPT-4o. Our experiments reveal that RLMs generate up to 70x more tokens than necessary, often failing at tasks that simpler non-reasoning models handle efficiently with higher accuracy. Our findings underscore the need for more effective training and inference strategies in RLMs.
Authors: Khoa Tran, Bao Huynh, Tri Le, Lam Pham, Vy-Rin Nguyen, Hung-Cuong Trinh, Duong Tran Anh
Abstract: Accurate prediction of the Remaining Useful Life (RUL) in Lithium ion battery (LIB) health management systems is essential for ensuring operational reliability and safety. However, many existing methods assume that training and testing data follow the same distribution, limiting their ability to generalize to unseen target domains. To address this, we propose a novel RUL prediction framework that incorporates a domain adaptation (DA) technique. Our framework integrates a signal preprocessing pipeline including noise reduction, feature extraction, and normalization with a robust deep learning model called HybridoNet Adapt. The model features a combination of LSTM, Multihead Attention, and Neural ODE layers for feature extraction, followed by two predictor modules with trainable trade-off parameters. To improve generalization, we adopt a DA strategy inspired by Domain Adversarial Neural Networks (DANN), replacing adversarial loss with Maximum Mean Discrepancy (MMD) to learn domain-invariant features. Experimental results show that HybridoNet Adapt significantly outperforms traditional models such as XGBoost and Elastic Net, as well as deep learning baselines like Dual input DNN, demonstrating its potential for scalable and reliable battery health management (BHM).
Authors: Tianren Zhang, Dai-Bei Yang
Abstract: Contemporary large language models (LLMs), such as GPT-4 and Llama, have harnessed extensive computational power and diverse text corpora to achieve remarkable proficiency in interpreting and generating domain-specific content, including materials science. To leverage the domain knowledge embedded within these models, we propose a simple yet effective multimodal architecture, PolyLLMem, which integrates text embeddings generated by Llama 3 with molecular structure embeddings derived from Uni-Mol, for polymer properties prediction tasks. In our model, Low-rank adaptation (LoRA) layers were also incorporated during the property prediction tasks to refine the embeddings based on our limited polymer dataset, thereby enhancing their chemical relevance for polymer SMILES representation. This balanced fusion of fine-tuned textual and structural information enables PolyLLMem to accurately predict a variety of polymer properties despite the scarcity of training data. Its performance is comparable to, and in some cases exceeds, that of graph-based models, as well as transformer-based models that typically require pretraining on millions of polymer samples. These findings demonstrate that LLM, such as Llama, can effectively capture chemical information encoded in polymer PSMILES, and underscore the efficacy of multimodal fusion of LLM embeddings and molecular structure embeddings in overcoming data scarcity and accelerating the discovery of advanced polymeric materials.
Authors: Fabrizio Garuti, Enver Sangineto, Simone Luetto, Lorenzo Forni, Rita Cucchiara
Abstract: Tabular data generation has recently attracted a growing interest due to its different application scenarios. However, generating time series of tabular data, where each element of the series depends on the others, remains a largely unexplored domain. This gap is probably due to the difficulty of jointly solving different problems, the main of which are the heterogeneity of tabular data (a problem common to non-time-dependent approaches) and the variable length of a time series. In this paper, we propose a Diffusion Transformers (DiTs) based approach for tabular data series generation. Inspired by the recent success of DiTs in image and video generation, we extend this framework to deal with heterogeneous data and variable-length sequences. Using extensive experiments on six datasets, we show that the proposed approach outperforms previous work by a large margin.
Authors: Jiaqi He, Xiangwen Luo, Yiping Wang
Abstract: At the current stage, deep learning-based methods have demonstrated excellent capabilities in evaluating aerodynamic performance, significantly reducing the time and cost required for traditional computational fluid dynamics (CFD) simulations. However, when faced with the task of processing extremely complex three-dimensional (3D) vehicle models, the lack of large-scale datasets and training resources, coupled with the inherent diversity and complexity of the geometry of different vehicle models, means that the prediction accuracy and versatility of these networks are still not up to the level required for current production. In view of the remarkable success of Transformer models in the field of natural language processing and their strong potential in the field of image processing, this study innovatively proposes a point cloud learning framework called DrivAer Transformer (DAT). The DAT structure uses the DrivAerNet++ dataset, which contains high-fidelity CFD data of industrial-standard 3D vehicle shapes. enabling accurate estimation of air drag directly from 3D meshes, thus avoiding the limitations of traditional methods such as 2D image rendering or signed distance fields (SDF). DAT enables fast and accurate drag prediction, driving the evolution of the aerodynamic evaluation process and laying the critical foundation for introducing a data-driven approach to automotive design. The framework is expected to accelerate the vehicle design process and improve development efficiency.
Authors: Hongkang Li, Yihua Zhang, Shuai Zhang, Meng Wang, Sijia Liu, Pin-Yu Chen
Abstract: Task arithmetic refers to editing the pre-trained model by adding a weighted sum of task vectors, each of which is the weight update from the pre-trained model to fine-tuned models for certain tasks. This approach recently gained attention as a computationally efficient inference method for model editing, e.g., multi-task learning, forgetting, and out-of-domain generalization capabilities. However, the theoretical understanding of why task vectors can execute various conceptual operations remains limited, due to the highly non-convexity of training Transformer-based models. To the best of our knowledge, this paper provides the first theoretical characterization of the generalization guarantees of task vector methods on nonlinear Transformers. We consider a conceptual learning setting, where each task is a binary classification problem based on a discriminative pattern. We theoretically prove the effectiveness of task addition in simultaneously learning a set of irrelevant or aligned tasks, as well as the success of task negation in unlearning one task from irrelevant or contradictory tasks. Moreover, we prove the proper selection of linear coefficients for task arithmetic to achieve guaranteed generalization to out-of-domain tasks. All of our theoretical results hold for both dense-weight parameters and their low-rank approximations. Although established in a conceptual setting, our theoretical findings were validated on a practical machine unlearning task using the large language model Phi-1.5 (1.3B).
Authors: Hongliang Lu, Shuqi Shen, Junjie Yang, Chao Lu, Xinhu Zheng, Hai Yang
Abstract: More than the adherence to specific traffic regulations, driving culture touches upon a more implicit part - an informal, conventional, collective behavioral pattern followed by drivers - that varies across countries, regions, and even cities. Such cultural divergence has become one of the biggest challenges in deploying autonomous vehicles (AVs) across diverse regions today. The current emergence of data-driven methods has shown a potential solution to enable culture-compatible driving through learning from data, but what if some underdeveloped regions cannot provide sufficient local data to inform driving culture? This issue is particularly significant for a broader global AV market. Here, we propose a cross-cultural deployment scheme for AVs, called data-light inverse reinforcement learning, designed to re-calibrate culture-specific AVs and assimilate them into other cultures. First, we report the divergence in driving cultures through a comprehensive comparative analysis of naturalistic driving datasets on highways from three countries: Germany, China, and the USA. Then, we demonstrate the effectiveness of our scheme by testing the expeditious cross-cultural deployment across these three countries, with cumulative testing mileage of over 56084 km. The performance is particularly advantageous when cross-cultural deployment is carried out without affluent local data. Results show that we can reduce the dependence on local data by a margin of 98.67% at best. This study is expected to bring a broader, fairer AV global market, particularly in those regions that lack enough local data to develop culture-compatible AVs.
Authors: Aaron Havens, Benjamin Kurt Miller, Bing Yan, Carles Domingo-Enrich, Anuroop Sriram, Brandon Wood, Daniel Levine, Bin Hu, Brandon Amos, Brian Karrer, Xiang Fu, Guan-Horng Liu, Ricky T. Q. Chen
Abstract: We introduce Adjoint Sampling, a highly scalable and efficient algorithm for learning diffusion processes that sample from unnormalized densities, or energy functions. It is the first on-policy approach that allows significantly more gradient updates than the number of energy evaluations and model samples, allowing us to scale to much larger problem settings than previously explored by similar methods. Our framework is theoretically grounded in stochastic optimal control and shares the same theoretical guarantees as Adjoint Matching, being able to train without the need for corrective measures that push samples towards the target distribution. We show how to incorporate key symmetries, as well as periodic boundary conditions, for modeling molecules in both cartesian and torsional coordinates. We demonstrate the effectiveness of our approach through extensive experiments on classical energy functions, and further scale up to neural network-based energy models where we perform amortized conformer generation across many molecular systems. To encourage further research in developing highly scalable sampling methods, we plan to open source these challenging benchmarks, where successful methods can directly impact progress in computational chemistry.
Authors: Ashwinee Panda, Vatsal Baherwani, Zain Sarwar, Benjamin Therien, Supriyo Chakraborty, Tom Goldstein
Abstract: Mixture of Experts (MoE) pretraining is more scalable than dense Transformer pretraining, because MoEs learn to route inputs to a sparse set of their feedforward parameters. However, this means that MoEs only receive a sparse backward update, leading to training instability and suboptimal performance. We present a lightweight approximation method that gives the MoE router a dense gradient update while continuing to sparsely activate its parameters. Our method, which we refer to as Default MoE, substitutes missing expert activations with default outputs consisting of an exponential moving average of expert outputs previously seen over the course of training. This allows the router to receive signals from every expert for each token, leading to significant improvements in training performance. Our Default MoE outperforms standard TopK routing in a variety of settings without requiring significant computational overhead. Code: https://github.com/vatsal0/default-moe.
Authors: Vishrut Ramraj, Nithin Nagaraj, Harikrishnan N B
Abstract: In this paper, we explore the application of Permutation Decision Trees (PDT) and strategic trailing for predicting stock market movements and executing profitable trades in the Indian stock market. We focus on high-frequency data using 5-minute candlesticks for the top 50 stocks listed in the NIFTY 50 index. We implement a trading strategy that aims to buy stocks at lower prices and sell them at higher prices, capitalizing on short-term market fluctuations. Due to regulatory constraints in India, short selling is not considered in our strategy. The model incorporates various technical indicators and employs hyperparameters such as the trailing stop-loss value and support thresholds to manage risk effectively. Our results indicate that the proposed trading bot has the potential to outperform the market average and yield returns higher than the risk-free rate offered by 10-year Indian government bonds. We trained and tested data on a 60 day dataset provided by Yahoo Finance. Specifically, 12 days for testing and 48 days for training. Our bot based on permutation decision tree achieved a profit of 1.3468 % over a 12-day testing period, where as a bot based on LSTM gave a return of 0.1238 % over a 12-day testing period and a bot based on RNN gave a return of 0.3096 % over a 12-day testing period. All of the bots outperform the buy-and-hold strategy, which resulted in a loss of 2.2508 %.
Authors: Shiwen Qin, Gabriela Kadlecov\'a, Martin Pil\'at, Shay B. Cohen, Roman Neruda, Elliot J. Crowley, Jovita Lukasik, Linus Ericsson
Abstract: Neural architecture search (NAS) faces a challenge in balancing the exploration of expressive, broad search spaces that enable architectural innovation with the need for efficient evaluation of architectures to effectively search such spaces. We investigate surrogate model training for improving search in highly expressive NAS search spaces based on context-free grammars. We show that i) surrogate models trained either using zero-cost-proxy metrics and neural graph features (GRAF) or by fine-tuning an off-the-shelf LM have high predictive power for the performance of architectures both within and across datasets, ii) these surrogates can be used to filter out bad architectures when searching on novel datasets, thereby significantly speeding up search and achieving better final performances, and iii) the surrogates can be further used directly as the search objective for huge speed-ups.
Authors: Andrew Kiruluta, Samantha Williams
Abstract: This paper presents a Sparse Hierarchical Fourier Interaction Networks, an architectural building block that unifies three complementary principles of frequency domain modeling: A hierarchical patch wise Fourier transform that affords simultaneous access to local detail and global context; A learnable, differentiable top K masking mechanism which retains only the most informative spectral coefficients, thereby exploiting the natural compressibility of visual and linguistic signals.
Authors: Antonio Emanuele Cin\`a, Ambra Demontis, Battista Biggio, Fabio Roli, Marcello Pelillo
Abstract: Sponge examples are test-time inputs optimized to increase energy consumption and prediction latency of deep networks deployed on hardware accelerators. By increasing the fraction of neurons activated during classification, these attacks reduce sparsity in network activation patterns, worsening the performance of hardware accelerators. In this work, we present a novel training-time attack, named sponge poisoning, which aims to worsen energy consumption and prediction latency of neural networks on any test input without affecting classification accuracy. To stage this attack, we assume that the attacker can control only a few model updates during training -- a likely scenario, e.g., when model training is outsourced to an untrusted third party or distributed via federated learning. Our extensive experiments on image classification tasks show that sponge poisoning is effective, and that fine-tuning poisoned models to repair them poses prohibitive costs for most users, highlighting that tackling sponge poisoning remains an open issue.
Authors: Junyu Liu, Frederik Wilde, Antonio Anna Mele, Xin Jin, Liang Jiang, Jens Eisert
Abstract: Saddle points constitute a crucial challenge for first-order gradient descent algorithms. In notions of classical machine learning, they are avoided for example by means of stochastic gradient descent methods. In this work, we provide evidence that the saddle points problem can be naturally avoided in variational quantum algorithms by exploiting the presence of stochasticity. We prove convergence guarantees and present practical examples in numerical simulations and on quantum hardware. We argue that the natural stochasticity of variational algorithms can be beneficial for avoiding strict saddle points, i.e., those saddle points with at least one negative Hessian eigenvalue. This insight that some levels of shot noise could help is expected to add a new perspective to notions of near-term variational quantum algorithms.
Authors: Hristos Tyralis, Georgia Papacharalampous, Nilay Dogulu, Kwok P. Chun
Abstract: Typical machine learning regression applications aim to report the mean or the median of the predictive probability distribution, via training with a squared or an absolute error scoring function. The importance of issuing predictions of more functionals of the predictive probability distribution (quantiles and expectiles) has been recognized as a means to quantify the uncertainty of the prediction. In deep learning (DL) applications, that is possible through quantile and expectile regression neural networks (QRNN and ERNN respectively). Here we introduce deep Huber quantile regression networks (DHQRN) that nest QRNN and ERNN as edge cases. DHQRN can predict Huber quantiles, which are more general functionals in the sense that they nest quantiles and expectiles as limiting cases. The main idea is to train a DL algorithm with the Huber quantile scoring function, which is consistent for the Huber quantile functional. As a proof of concept, DHQRN are applied to predict house prices in Melbourne, Australia and Boston, United States (US). In this context, predictive performances of three DL architectures are discussed along with evidential interpretation of results from two economic case studies. Additional simulation experiments and applications to real-world case studies using open datasets demonstrate a satisfactory absolute performance of DHQRN.
Authors: Shizhan Lu, Zeshui Xu, Zhu Fu, Longsheng Cheng, Tongbin Yang
Abstract: Hesitant fuzzy sets find extensive application in specific scenarios involving uncertainty and hesitation. In the context of set theory, the concept of inclusion relationship holds significant importance as a fundamental definition. Consequently, as a type of sets, hesitant fuzzy sets necessitate a clear and explicit definition of the inclusion relationship. Based on the discrete form of hesitant fuzzy membership degrees, this study proposes multiple types of inclusion relationships for hesitant fuzzy sets. Subsequently, this paper introduces foundational propositions related to hesitant fuzzy sets, as well as propositions concerning families of hesitant fuzzy sets.
Authors: Yiheng Jiang, Sinho Chewi, Aram-Alexandre Pooladian
Abstract: We develop a theory of finite-dimensional polyhedral subsets over the Wasserstein space and optimization of functionals over them via first-order methods. Our main application is to the problem of mean-field variational inference, which seeks to approximate a distribution $\pi$ over $\mathbb{R}^d$ by a product measure $\pi^\star$. When $\pi$ is strongly log-concave and log-smooth, we provide (1) approximation rates certifying that $\pi^\star$ is close to the minimizer $\pi^\star_\diamond$ of the KL divergence over a \emph{polyhedral} set $\mathcal{P}_\diamond$, and (2) an algorithm for minimizing $\text{KL}(\cdot\|\pi)$ over $\mathcal{P}_\diamond$ based on accelerated gradient descent over $\R^d$. As a byproduct of our analysis, we obtain the first end-to-end analysis for gradient-based algorithms for MFVI.
Authors: Maksym Andriushchenko, Francesco Croce, Nicolas Flammarion
Abstract: We show that even the most recent safety-aligned LLMs are not robust to simple adaptive jailbreaking attacks. First, we demonstrate how to successfully leverage access to logprobs for jailbreaking: we initially design an adversarial prompt template (sometimes adapted to the target LLM), and then we apply random search on a suffix to maximize a target logprob (e.g., of the token "Sure"), potentially with multiple restarts. In this way, we achieve 100% attack success rate -- according to GPT-4 as a judge -- on Vicuna-13B, Mistral-7B, Phi-3-Mini, Nemotron-4-340B, Llama-2-Chat-7B/13B/70B, Llama-3-Instruct-8B, Gemma-7B, GPT-3.5, GPT-4o, and R2D2 from HarmBench that was adversarially trained against the GCG attack. We also show how to jailbreak all Claude models -- that do not expose logprobs -- via either a transfer or prefilling attack with a 100% success rate. In addition, we show how to use random search on a restricted set of tokens for finding trojan strings in poisoned models -- a task that shares many similarities with jailbreaking -- which is the algorithm that brought us the first place in the SaTML'24 Trojan Detection Competition. The common theme behind these attacks is that adaptivity is crucial: different models are vulnerable to different prompting templates (e.g., R2D2 is very sensitive to in-context learning prompts), some models have unique vulnerabilities based on their APIs (e.g., prefilling for Claude), and in some settings, it is crucial to restrict the token search space based on prior knowledge (e.g., for trojan detection). For reproducibility purposes, we provide the code, logs, and jailbreak artifacts in the JailbreakBench format at https://github.com/tml-epfl/llm-adaptive-attacks.
Authors: Hao Zhao, Maksym Andriushchenko, Francesco Croce, Nicolas Flammarion
Abstract: In-context learning (ICL) allows LLMs to learn from examples without changing their weights: this is a particularly promising capability for long-context LLMs that can potentially learn from many examples. Recently, Lin et al. (2024) proposed URIAL, a method using only three in-context examples to align base LLMs, achieving non-trivial instruction following performance. In this work, we show that, while effective, ICL alignment with URIAL still underperforms compared to instruction fine-tuning on the established benchmark MT-Bench, especially with more capable base LLMs. We then uncover the most relevant elements for successful in-context alignment, finding the crucial role of the decoding parameters. Based on these insights, we show that the approach of URIAL can indeed be improved by adding high-quality, potentially carefully selected via greedy search, demonstrations in context, getting closer to the performance of instruct models. Finally, we provide the first, to our knowledge, systematic comparison of ICL and instruction fine-tuning (IFT) for instruction following in the low data regime, where ICL can be a viable alternative to IFT. Overall, our work advances the understanding of ICL as an alignment technique and its relationship to IFT. We provide our code at https://github.com/tml-epfl/icl-alignment.
Authors: Tianyu Zhang, Suyuchen Wang, Lu Li, Ge Zhang, Perouz Taslakian, Sai Rajeswar, Jie Fu, Bang Liu, Yoshua Bengio
Abstract: We introduce Visual Caption Restoration (VCR), a novel vision-language task that challenges models to accurately restore partially obscured texts using pixel-level hints within images. This task stems from the observation that text embedded in images is intrinsically different from common visual elements and natural language due to the need to align the modalities of vision, text, and text embedded in images. While numerous works have integrated text embedded in images into visual question-answering tasks, approaches to these tasks generally rely on optical character recognition or masked language modeling, thus reducing the task to mainly text-based processing. However, text-based processing becomes ineffective in VCR as accurate text restoration depends on the combined information from provided images, context, and subtle cues from the tiny exposed areas of masked texts. We develop a pipeline to generate synthetic images for the VCR task using image-caption pairs, with adjustable caption visibility to control the task difficulty. With this pipeline, we construct a dataset for VCR called VCR-Wiki using images with captions from Wikipedia, comprising 2.11M English and 346K Chinese entities in both easy and hard split variants. Our results reveal that current vision language models significantly lag behind human performance in the VCR task, and merely fine-tuning the models on our dataset does not lead to notable improvements. We release VCR-Wiki and the data construction code to facilitate future research.
Authors: Rachmad Vidya Wicaksana Putra, Muhammad Shafique
Abstract: Intelligent mobile agents (e.g., UGVs and UAVs) typically demand low power/energy consumption when solving their machine learning (ML)-based tasks, since they are usually powered by portable batteries with limited capacity. A potential solution is employing neuromorphic computing with Spiking Neural Networks (SNNs), which leverages event-based computation to enable ultra-low power/energy ML algorithms. To maximize the performance efficiency of SNN inference, the In-Memory Computing (IMC)-based hardware accelerators with emerging device technologies (e.g., RRAM) can be employed. However, SNN models are typically developed without considering constraints from the application and the underlying IMC hardware, thereby hindering SNNs from reaching their full potential in performance and efficiency. To address this, we propose NeuroNAS, a novel framework for developing energyefficient neuromorphic IMC for intelligent mobile agents using hardware-aware spiking neural architecture search (NAS), i.e., by quickly finding an SNN architecture that offers high accuracy under the given constraints (e.g., memory, area, latency, and energy consumption). Its key steps include: optimizing SNN operations to enable efficient NAS, employing quantization to minimize the memory footprint, developing an SNN architecture that facilitates an effective learning, and devising a systematic hardware-aware search algorithm to meet the constraints. Compared to the state-of-the-art techniques, NeuroNAS quickly finds SNN architectures (with 8bit weight precision) that maintain high accuracy by up to 6.6x search time speed-ups, while achieving up to 92% area savings, 1.2x latency improvements, 84% energy savings across different datasets (i.e., CIFAR-10, CIFAR-100, and TinyImageNet-200); while the state-of-the-art fail to meet all constraints at once.
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, o1-mini, o1-preview, 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: Kensen Shi, Deniz Alt{\i}nb\"uken, Saswat Anand, Mihai Christodorescu, Katja Gr\"unwedel, Alexa Koenings, Sai Naidu, Anurag Pathak, Marc Rasi, Fredde Ribeiro, Brandon Ruffin, Siddhant Sanyam, Maxim Tabachnyk, Sara Toth, Roy Tu, Tobias Welp, Pengcheng Yin, Manzil Zaheer, Satish Chandra, Charles Sutton
Abstract: We propose using natural language outlines as a novel modality and interaction surface for providing AI assistance to developers throughout the software development process. An NL outline for a code function comprises multiple statements written in concise prose, which partition the code and summarize its main ideas in the style of literate programming. Crucially, we find that modern LLMs can generate accurate and high-quality NL outlines in practice. Moreover, NL outlines enable a bidirectional sync between code and NL, where a developer can change either code or NL and have the LLM automatically update the other. We discuss many use cases for NL outlines: they can accelerate understanding and navigation of code and diffs, simplify code maintenance, augment code search, steer code generation, and more. We then propose and compare multiple LLM prompting techniques for generating outlines and ask professional developers to judge outline quality. Finally, we present two case studies applying NL outlines toward code review and malware detection.
Authors: Landon Dyken, Saugat Adhikari, Pravin Poudel, Steve Petruzza, Da Yan, Will Usher, Sidharth Kumar
Abstract: Mapping the extent of flood events is a necessary and important aspect of disaster management. In recent years, deep learning methods have evolved as an effective tool to quickly label high resolution imagery and provide necessary flood extent mappings. These methods, though, require large amounts of annotated training data to create models that are accurate and robust to new flooded imagery. In this work, we present FloodTrace, a web-based application that enables effective crowdsourcing of flooded region annotation for machine learning applications. To create this application, we conducted extensive interviews with domain experts to produce a set of formal requirements. Our work brings topological segmentation tools to the web and greatly improves annotation efficiency compared to the state-of-the-art. The user-friendliness of our solution allows researchers to outsource annotations to non-experts and utilize them to produce training data with equal quality to fully expert-labeled data. We conducted a user study to confirm the effectiveness of our application in which 266 graduate students annotated high-resolution aerial imagery from Hurricane Matthew in North Carolina. Experimental results show the efficiency benefits of our application for untrained users, with median annotation time less than half the state-of-the-art annotation method. In addition, using our aggregation and correction framework, flood detection models trained on crowdsourced annotations were able to achieve performance equal to models trained on fully expert-labeled annotations, while requiring a fraction of the time on the part of the expert.
Authors: Yijia Xiao, Edward Sun, Yiqiao Jin, Qifan Wang, Wei Wang
Abstract: Understanding biological processes, drug development, and biotechnological advancements requires a detailed analysis of protein structures and functions, a task that is inherently complex and time-consuming in traditional protein research. To streamline this process, we introduce ProteinGPT, a state-of-the-art multimodal large language model for proteins that enables users to upload protein sequences and/or structures for comprehensive analysis and responsive inquiries. ProteinGPT integrates protein sequence and structure encoders with linear projection layers to ensure precise representation adaptation and leverages a large language model (LLM) to generate accurate, contextually relevant responses. To train ProteinGPT, we constructed a large-scale dataset of 132,092 proteins, each annotated with 20-30 property tags and 5-10 QA pairs per protein, and optimized the instruction-tuning process using GPT-4o. Experiments demonstrate that ProteinGPT effectively generates informative responses to protein-related questions, achieving high performance on both semantic and lexical metrics and significantly outperforming baseline models and general-purpose LLMs in understanding and responding to protein-related queries. Our code and data are available at https://github.com/ProteinGPT/ProteinGPT.
Authors: Ashwin Samudre, Mircea Petrache, Brian D. Nord, Shubhendu Trivedi
Abstract: There has been much recent interest in designing neural networks (NNs) with relaxed equivariance, which interpolate between exact equivariance and full flexibility for consistent performance gains. In a separate line of work, structured parameter matrices with low displacement rank (LDR) -- which permit fast function and gradient evaluation -- have been used to create compact NNs, though primarily benefiting classical convolutional neural networks (CNNs). In this work, we propose a framework based on symmetry-based structured matrices to build approximately equivariant NNs with fewer parameters. Our approach unifies the aforementioned areas using Group Matrices (GMs), a forgotten precursor to the modern notion of regular representations of finite groups. GMs allow the design of structured matrices similar to LDR matrices, which can generalize all the elementary operations of a CNN from cyclic groups to arbitrary finite groups. We show GMs can also generalize classical LDR theory to general discrete groups, enabling a natural formalism for approximate equivariance. We test GM-based architectures on various tasks with relaxed symmetry and find that our framework performs competitively with approximately equivariant NNs and other structured matrix-based methods, often with one to two orders of magnitude fewer parameters.
Authors: Julia S. Schmid (Department of Mathematical and Statistical Sciences, University of Alberta, Edmonton, Alberta, Canada), Sean Simmons (Anglers Atlas, Goldstream Publishing, Prince George, British Columbia, Canada), Mark A. Lewis (Department of Mathematical and Statistical Sciences, University of Alberta, Edmonton, Alberta, Canada, Department of Mathematics and Statistics, University of Victoria, Victoria, British Columbia, Canada, Department of Biology, University of Victoria, Victoria, British Columbia, Canada, Department of Biological Sciences, University of Alberta, Edmonton, Alberta, Canada), Mark S. Poesch (Department of Biological Sciences, University of Alberta, Edmonton, Alberta, Canada), Pouria Ramazi (Department of Mathematics and Statistics, Brock University, St. Catharines, Ontario, Canada)
Abstract: Understanding and predicting recreational angler effort is important for sustainable fisheries management. However, conventional methods of measuring angler effort, such as surveys, can be costly and limited in both time and spatial extent. Models that predict angler effort based on environmental or economic factors typically rely on historical data, which often limits their spatial and temporal generalizability due to data scarcity. In this study, high-resolution data from an online fishing platform and easily accessible auxiliary data were tested to predict daily boat presence and aerial counts of boats at almost 200 lakes over five years in Ontario, Canada. Lake-information website visits alone enabled predicting daily angler boat presence with 78% accuracy. While incorporating additional environmental, socio-ecological, weather and angler-reported features into machine learning models did not remarkably improve prediction performance of boat presence, they were substantial for the prediction of boat counts. Models achieved an R2 of up to 0.77 at known lakes included in the model training, but they performed poorly for unknown lakes (R2 = 0.21). The results demonstrate the value of integrating data from online fishing platforms into predictive models and highlight the potential of machine learning models to enhance fisheries management.
Authors: Michele Caprio
Abstract: We present generalized versions of Monge's and Kantorovich's optimal transport problems with the probabilities being transported replaced by lower probabilities. We show that, when the lower probabilities are the lower envelopes of $\epsilon$-contaminated sets, then our version of Monge's, and a restricted version of our Kantorovich's problems, coincide with their respective classical versions. We also give sufficient conditions for the existence of our version of Kantorovich's optimal plan, and for the two problems to be equivalent. As a byproduct, we show that for $\epsilon$-contaminations the lower probability versions of Monge's and Kantorovich's optimal transport problems need not coincide. The applications of our results to Machine Learning and Artificial Intelligence are also discussed.
Authors: Santhosh Kumar Ramakrishnan, Erik Wijmans, Philipp Kraehenbuehl, Vladlen Koltun
Abstract: Not yet. We present SPACE, a benchmark that systematically evaluates spatial cognition in frontier models. Our benchmark builds on decades of research in cognitive science. It evaluates large-scale mapping abilities that are brought to bear when an organism traverses physical environments, smaller-scale reasoning about object shapes and layouts, and cognitive infrastructure such as spatial attention and memory. For many tasks, we instantiate parallel presentations via text and images, allowing us to benchmark both large language models and large multimodal models. Results suggest that contemporary frontier models fall short of the spatial intelligence of animals, performing near chance level on a number of classic tests of animal cognition. Code and data are available: https://github.com/apple/ml-space-benchmark
Authors: David Yunis, Siyu Huo, Chulaka Gunasekara, Danish Contractor
Abstract: We now deploy language models in a wide variety of user-facing applications. Typically, these deployments have some specific purpose, like answering questions about documentation or acting as coding assistants, but they require general language understanding. Under these circumstances these models should not be able to answer irrelevant requests such as, poetry generation or questions about physics, etc. Instead we would like language models to only answer to queries corresponding to desired behavior and refuse all other requests, which we refer to as scoping. We conduct a comprehensive empirical evaluation of potential methods from prompting to fine-tuning to preference learning to a recently proposed method for general alignment called Circuit Breakers (CB). Across three families of language models and a broad variety of tasks, we show that it is possible to scope language models. We examine scoping for multiple topics, and fine-grained topics. We ablate diversity of irrelevant queries, layer different techniques, conduct adversarial evaluations and more. Among other results, we find that, when diverse examples of irrelevant queries are available, simple supervised fine-tuning produces the best results, but when such diversity is low, Circuit Breakers perform quite well. One can often get the benefits of both methods by layering them in succession. We intend our study to serve as a practitioner's guide to scoping language models.
Authors: Laurin Lux, Alexander H. Berger, Alexander Weers, Nico Stucki, Daniel Rueckert, Ulrich Bauer, Johannes C. Paetzold
Abstract: Topological correctness plays a critical role in many image segmentation tasks, yet most networks are trained using pixel-wise loss functions, such as Dice, neglecting topological accuracy. Existing topology-aware methods often lack robust topological guarantees, are limited to specific use cases, or impose high computational costs. In this work, we propose a novel, graph-based framework for topologically accurate image segmentation that is both computationally efficient and generally applicable. Our method constructs a component graph that fully encodes the topological information of both the prediction and ground truth, allowing us to efficiently identify topologically critical regions and aggregate a loss based on local neighborhood information. Furthermore, we introduce a strict topological metric capturing the homotopy equivalence between the union and intersection of prediction-label pairs. We formally prove the topological guarantees of our approach and empirically validate its effectiveness on binary and multi-class datasets. Our loss demonstrates state-of-the-art performance with up to fivefold faster loss computation compared to persistent homology methods.
Authors: Ziyao Zeng, Jingcheng Ni, Daniel Wang, Patrick Rim, Younjoon Chung, Fengyu Yang, Byung-Woo Hong, Alex Wong
Abstract: Traditional monocular depth estimation suffers from inherent ambiguity and visual nuisance. We argue that language prior can enhance monocular depth estimation by leveraging the inductive bias learned during the text-to-image pre-training of diffusion models. The ability of these models to generate images that align with text indicates that they have learned the spatial relationships, size, and shape of specified objects, which can be applied to improve depth estimation. Thus, we propose PriorDiffusion, using a pre-trained text-to-image diffusion model that takes both images and corresponding text descriptions to infer affine-invariant depth through a denoising process. We also show that language prior enhances the model's perception of specific regions of images that users care about and describe. Simultaneously, language prior acts as a constraint to accelerate the convergence of both training and the inference diffusion trajectory. By training on HyperSim and Virtual KITTI, we achieve faster training convergence, fewer inference diffusion steps, and state-of-the-art zero-shot performance across NYUv2, KITTI, ETH3D, and ScanNet. Code will be released upon acceptance.
Authors: Jungbin Cho, Junwan Kim, Jisoo Kim, Minseo Kim, Mingu Kang, Sungeun Hong, Tae-Hyun Oh, Youngjae Yu
Abstract: Human motion is inherently continuous and dynamic, posing significant challenges for generative models. While discrete generation methods are widely used, they suffer from limited expressiveness and frame-wise noise artifacts. In contrast, continuous approaches produce smoother, more natural motion but often struggle to adhere to conditioning signals due to high-dimensional complexity and limited training data. To resolve this discord between discrete and continuous representations, we introduce DisCoRD: Discrete Tokens to Continuous Motion via Rectified Flow Decoding, a novel method that leverages rectified flow to decode discrete motion tokens in the continuous, raw motion space. Our core idea is to frame token decoding as a conditional generation task, ensuring that DisCoRD captures fine-grained dynamics and achieves smoother, more natural motions. Compatible with any discrete-based framework, our method enhances naturalness without compromising faithfulness to the conditioning signals on diverse settings. Extensive evaluations Our project page is available at: https://whwjdqls.github.io/discord.github.io/.
Authors: Chenlu Ding, Jiancan Wu, Yancheng Yuan, Jinda Lu, Kai Zhang, Alex Su, Xiang Wang, Xiangnan He
Abstract: The advent of Large Language Models (LLMs) has revolutionized natural language processing, enabling advanced understanding and reasoning capabilities across a variety of tasks. Fine-tuning these models for specific domains, particularly through Parameter-Efficient Fine-Tuning (PEFT) strategies like LoRA, has become a prevalent practice due to its efficiency. However, this raises significant privacy and security concerns, as models may inadvertently retain and disseminate sensitive or undesirable information. To address these issues, we introduce a novel instance-wise unlearning framework, LLMEraser, which systematically categorizes unlearning tasks and applies precise parameter adjustments using influence functions. Unlike traditional unlearning techniques that are often limited in scope and require extensive retraining, LLMEraser is designed to handle a broad spectrum of unlearning tasks without compromising model performance. Extensive experiments on benchmark datasets demonstrate that LLMEraser excels in efficiently managing various unlearning scenarios while maintaining the overall integrity and efficacy of the models.
Authors: Francesco Villani, Igor Maljkovic, Dario Lazzaro, Angelo Sotgiu, Antonio Emanuele Cin\`a, Fabio Roli
Abstract: Deep Neural Networks are vulnerable to adversarial examples, i.e., carefully crafted input samples that can cause models to make incorrect predictions with high confidence. To mitigate these vulnerabilities, adversarial training and detection-based defenses have been proposed to strengthen models in advance. However, most of these approaches focus on a single data modality, overlooking the relationships between visual patterns and textual descriptions of the input. In this paper, we propose a novel defense, MultiShield, designed to combine and complement these defenses with multi-modal information to further enhance their robustness. MultiShield leverages multi-modal large language models to detect adversarial examples and abstain from uncertain classifications when there is no alignment between textual and visual representations of the input. Extensive evaluations on CIFAR-10 and ImageNet datasets, using robust and non-robust image classification models, demonstrate that MultiShield can be easily integrated to detect and reject adversarial examples, outperforming the original defenses.
Authors: Tassilo Wald, Constantin Ulrich, Jonathan Suprijadi, Sebastian Ziegler, Michal Nohel, Robin Peretzke, Gregor K\"ohler, Klaus H. Maier-Hein
Abstract: The field of self-supervised learning (SSL) for 3D medical images lacks consistency and standardization. While many methods have been developed, it is impossible to identify the current state-of-the-art, due to i) varying and small pretraining datasets, ii) varying architectures, and iii) being evaluated on differing downstream datasets. In this paper, we bring clarity to this field and lay the foundation for further method advancements through three key contributions: We a) publish the largest publicly available pre-training dataset comprising 114k 3D brain MRI volumes, enabling all practitioners to pre-train on a large-scale dataset. We b) benchmark existing 3D self-supervised learning methods on this dataset for a state-of-the-art CNN and Transformer architecture, clarifying the state of 3D SSL pre-training. Among many findings, we show that pre-trained methods can exceed a strong from-scratch nnU-Net ResEnc-L baseline. Lastly, we c) publish the code of our pre-training and fine-tuning frameworks and provide the pre-trained models created during the benchmarking process to facilitate rapid adoption and reproduction.
Authors: Michele Caprio, Yusuf Sale, Eyke H\"ullermeier
Abstract: Recently, Cella and Martin proved how, under an assumption called consonance, a credal set (i.e. a closed and convex set of probabilities) can be derived from the conformal transducer associated with transductive conformal prediction. We show that the Imprecise Highest Density Region (IHDR) associated with such a credal set corresponds to the classical Conformal Prediction Region. In proving this result, we establish a new relationship between Conformal Prediction and Imprecise Probability (IP) theories, via the IP concept of a cloud. A byproduct of our presentation is the discovery that consonant plausibility functions are monoid homomorphisms, a new algebraic property of an IP tool.
Authors: Michael Doherty, Alejandra Beghelli
Abstract: Many works have investigated reinforcement learning (RL) for routing and spectrum assignment on flex-grid networks but only one work to date has examined RL for fixed-grid with flex-rate transponders, despite production systems using this paradigm. Flex-rate transponders allow existing lightpaths to accommodate new services, a task we term routing and wavelength assignment with lightpath reuse (RWA-LR). We re-examine this problem and present a thorough benchmarking of heuristic algorithms for RWA-LR, which are shown to have 6% increased throughput when candidate paths are ordered by number of hops, rather than total length. We train an RL agent for RWA-LR with graph attention networks for the policy and value functions to exploit the graph-structured data. We provide details of our methodology and open source all of our code for reproduction. We outperform the previous state-of-the-art RL approach by 2.5% (17.4 Tbps mean additional throughput) and the best heuristic by 1.2% (8.5 Tbps mean additional throughput). This marginal gain highlights the difficulty in learning effective RL policies on long horizon resource allocation tasks.
Authors: Chengchun Shi
Abstract: Reinforcement learning (RL) is concerned with how intelligence agents take actions in a given environment to maximize the cumulative reward they receive. In healthcare, applying RL algorithms could assist patients in improving their health status. In ride-sharing platforms, applying RL algorithms could increase drivers' income and customer satisfaction. For large language models, applying RL algorithms could align their outputs with human preferences. Over the past decade, RL has been arguably one of the most vibrant research frontiers in machine learning. Nevertheless, statistics as a field, as opposed to computer science, has only recently begun to engage with RL both in depth and in breadth. This chapter presents a selective review of statistical inferential tools for RL, covering both hypothesis testing and confidence interval construction. Our goal is to highlight the value of statistical inference in RL for both the statistics and machine learning communities, and to promote the broader application of classical statistical inference tools in this vibrant area of research.
Authors: Hristos Tyralis, Georgia Papacharalampous
Abstract: Loss functions constructed by applying transformations to the realization and prediction variables of (strictly) consistent loss functions have been extensively studied empirically, yet their theoretical foundations remain unexplored. To address this gap, we establish formal characterizations of (strict) consistency for such transformed loss functions and their corresponding elicitable functionals. Our analysis focuses on two interrelated cases: (a) transformations applied solely to the realization variable and (b) bijective transformations applied jointly to both the realization and prediction variables. These cases extend the well-established framework of transformations applied exclusively to the prediction variable, as formalized by Osband's revelation principle. We further develop analogous characterizations for (strict) identification functions. The resulting theoretical framework is broadly applicable to statistical and machine learning methodologies. When applied to Bregman and expectile loss functions, our framework enables two key advancements: (a) the interpretation of empirical findings from models trained with transformed loss functions and (b) the systematic construction of novel identifiable and elicitable functionals, including the g-transformed expectation and g-transformed expectile. By unifying theoretical insights with practical applications, this work advances principled methodologies for designing loss functions in complex predictive tasks. Applications of the framework to simulated and real-world data illustrate its practical utility in diverse settings.
Authors: Dana Rubin, Allan dos Santos Costa, Manvitha Ponnapati, Joseph Jacobson
Abstract: Ribonucleic acid (RNA) plays fundamental roles in biological systems, from carrying genetic information to performing enzymatic function. Understanding and designing RNA can enable novel therapeutic application and biotechnological innovation. To enhance RNA design, in this paper we introduce RiboGen, the first deep learning model to simultaneously generate RNA sequence and all-atom 3D structure. RiboGen leverages the standard Flow Matching with Discrete Flow Matching in a multimodal data representation. RiboGen is based on Euclidean Equivariant neural networks for efficiently processing and learning three-dimensional geometry. Our experiments show that RiboGen can efficiently generate chemically plausible and self-consistent RNA samples, suggesting that co-generation of sequence and structure is a competitive approach for modeling RNA.
Authors: Asifullah Khan, Laiba Asmatullah, Anza Malik, Shahzaib Khan, Hamna Asif
Abstract: Self-supervised learning is a machine learning approach that generates implicit labels by learning underlined patterns and extracting discriminative features from unlabeled data without manual labelling. Contrastive learning introduces the concept of "positive" and "negative" samples, where positive pairs (e.g., variation of the same image/object) are brought together in the embedding space, and negative pairs (e.g., views from different images/objects) are pushed farther away. This methodology has shown significant improvements in image understanding and image text analysis without much reliance on labeled data. In this paper, we comprehensively discuss the terminologies, recent developments and applications of contrastive learning with respect to text-image models. Specifically, we provide an overview of the approaches of contrastive learning in text-image models in recent years. Secondly, we categorize the approaches based on different model structures. Thirdly, we further introduce and discuss the latest advances of the techniques used in the process such as pretext tasks for both images and text, architectural structures, and key trends. Lastly, we discuss the recent state-of-art applications of self-supervised contrastive learning Text-Image based models.
Authors: Vinicius Hernandes, Thomas Spriggs, Saqar Khaleefah, Eliska Greplova
Abstract: Neural quantum states (NQS) have emerged as a powerful tool for approximating quantum wavefunctions using deep learning. While these models achieve remarkable accuracy, understanding how they encode physical information remains an open challenge. In this work, we introduce adiabatic fine-tuning, a scheme that trains NQS across a phase diagram, leading to strongly correlated weight representations across different models. This correlation in weight space enables the detection of phase transitions in quantum systems by analyzing the trained network weights alone. We validate our approach on the transverse field Ising model and the J1-J2 Heisenberg model, demonstrating that phase transitions manifest as distinct structures in weight space. Our results establish a connection between physical phase transitions and the geometry of neural network parameters, opening new directions for the interpretability of machine learning models in physics.
Authors: Ruiqi Zhang, Jingfeng Wu, Licong Lin, Peter L. Bartlett
Abstract: We study $\textit{gradient descent}$ (GD) for logistic regression on linearly separable data with stepsizes that adapt to the current risk, scaled by a constant hyperparameter $\eta$. We show that after at most $1/\gamma^2$ burn-in steps, GD achieves a risk upper bounded by $\exp(-\Theta(\eta))$, where $\gamma$ is the margin of the dataset. As $\eta$ can be arbitrarily large, GD attains an arbitrarily small risk $\textit{immediately after the burn-in steps}$, though the risk evolution may be $\textit{non-monotonic}$. We further construct hard datasets with margin $\gamma$, where any batch (or online) first-order method requires $\Omega(1/\gamma^2)$ steps to find a linear separator. Thus, GD with large, adaptive stepsizes is $\textit{minimax optimal}$ among first-order batch methods. Notably, the classical $\textit{Perceptron}$ (Novikoff, 1962), a first-order online method, also achieves a step complexity of $1/\gamma^2$, matching GD even in constants. Finally, our GD analysis extends to a broad class of loss functions and certain two-layer networks.
Authors: Dayu Yang, Antoine Simoulin, Xin Qian, Xiaoyi Liu, Yuwei Cao, Zhaopu Teng, Grey Yang
Abstract: High-quality code documentation is crucial for software development especially in the era of AI. However, generating it automatically using Large Language Models (LLMs) remains challenging, as existing approaches often produce incomplete, unhelpful, or factually incorrect outputs. We introduce DocAgent, a novel multi-agent collaborative system using topological code processing for incremental context building. Specialized agents (Reader, Searcher, Writer, Verifier, Orchestrator) then collaboratively generate documentation. We also propose a multi-faceted evaluation framework assessing Completeness, Helpfulness, and Truthfulness. Comprehensive experiments show DocAgent significantly outperforms baselines consistently. Our ablation study confirms the vital role of the topological processing order. DocAgent offers a robust approach for reliable code documentation generation in complex and proprietary repositories.