Machine Learning
Coarse-Grained Boltzmann Generators
Weilong Chen, Bojun Zhao, Jan Eckwert et al. (+1 more)
Sampling
equilibrium molecular configurations from the Boltzmann distribution is a longstanding challenge. Boltzmann Generators (BGs) address this by combining exact-likelihood generative models with importance sampling, but practical scalability is limited. Meanwhile, coarse-grained surrogates enable the modeling of larger systems by reducing effective dimensionality, yet often lack a reweighting procedure required to ensure asymptotically correct statistics. In this work, we propose Coarse-Grained Boltzmann Generators (CG-BGs), a framework for reduced-order generative modeling with importance sampling in coarse-grained coordinate space. CG-BGs generate samples using a flow-based model and reweight them using a learned potential of mean force (PMF). We show that the PMF can be learned from rapidly converged trajectories via enhanced sampling force matching. Experiments demonstrate that CG-BGs capture solvent-mediated interactions in highly reduced representations while substantially reducing computational cost relative to atomistic BGs, providing a practical route toward
equilibrium sampling of larger molecular systems.
Read the full abstract ▾
All authors Weilong Chen, Bojun Zhao, Jan Eckwert, Julija Zavadlav
Machine Learning · Cross-listed from cs.AI
OmniSapiens: A Foundation Model for Social Behavior Processing via Heterogeneity-Aware Relative Policy Optimization
Keane Ong, Sabri Boughorbel, Luwei Xiao et al. (+9 more)
Socially intelligent AI systems must reason across diverse human behavioral tasks and generalize to new social contexts. However, behavioral data is inherently heterogeneous, comprising diverse modalities and prediction targets that produce uneven training signals across samples, creating imbalanced learning dynamics that challenge existing AI models. To address this, we develop Omnisapiens-7B 2.0, a foundation model for social behavior processing that explicitly addresses learning from heterogeneous behavioral data. This is enabled through Heterogeneity-Aware Relative Policy Optimization, a new RL method that rebalances learning signals across samples by approximating each sample's contribution to the policy update and using these estimates to drive geometrically centered, inertially smoothed advantage modulation for stable training. Omnisapiens-7B 2.0 achieves the best and most consistent performance across 10 behavioral tasks, while also attaining the best performance on all five held-out
benchmarks, with gains of up to +12.02% and +9.37% respectively. Furthermore, it demonstrates more consistent and interpretable
reasoning traces, supporting reliable real-world behavioral applications. Our model is available at https://github.com/MIT-MI/human_behavior_atlas.
Read the full abstract ▾
All authors Keane Ong, Sabri Boughorbel, Luwei Xiao, Chanakya Ekbote, Wei Dai, Ao Qu, Jingyao Wu, Rui Mao, Ehsan Hoque, Erik Cambria, Gianmarco Mengaldo, Paul Pu Liang
Machine Learning
Generative clinical time series models trained on moderate amounts of patient data are privacy preserving
Rustam Zhumagambetov, Niklas Giesa, Sebastian D. Boie et al. (+1 more)
Sharing medical data for machine learning model training purposes is often impossible due to the risk of disclosing identifying information about individual patients. Synthetic data produced by generative artificial intelligence (genAI) models trained on real data is often seen as one possible solution to comply with
privacy regulations. While powerful genAI models for heterogeneous hospital time series have recently been introduced, such modeling does not guarantee
privacy protection, as the generated data may still reveal identifying information about individuals in the models' training cohort. Applying established
privacy mechanisms to generative time series models, however, proves challenging as post-hoc data anonymization through k-anonymization or similar techniques is limited, while model-centered
privacy mechanisms that implement differential
privacy (DP) may lead to unstable training, compromising the utility of generated data. Given these known limitations,
privacy audits for generative time series models are currently indispensable regardless of the concrete
privacy mechanisms applied to models and/or data. In this work, we use a battery of established
privacy attacks to audit state-of-the-art hospital time series models, trained on the public MIMIC-IV
dataset, with respect to
privacy preservation. Furthermore, the eICU
dataset was used to mount a
privacy attack against the synthetic data generator trained on the MIMIC-IV
dataset. Results show that established
privacy attacks are ineffective against generated multivariate clinical time series when synthetic data generators are trained on large enough training
datasets. Furthermore, we discuss how the use of existing DP mechanisms for these synthetic data generators would not bring desired improvement in
privacy, but only a decrease in utility for machine learning prediction tasks.
Read the full abstract ▾
All authors Rustam Zhumagambetov, Niklas Giesa, Sebastian D. Boie, Stefan Haufe
Machine Learning
Mitigating Reward Hacking in RLHF via Bayesian Non-negative Reward Modeling
Zhibin Duan, Guowei Rong, Zhuo Li et al. (+3 more)
Reward models learned from human preferences are central to aligning large
language models (LLMs) via reinforcement learning from human feedback, yet they are often vulnerable to reward hacking due to noisy annotations and systematic biases such as response length or style. We propose
Bayesian Non-Negative Reward Model (BNRM), a principled reward modeling framework that integrates non-negative factor analysis into Bradley-Terry (BT) preference model. BNRM represents rewards through a sparse, non-negative latent factor generative process that operates at two complementary levels: instance-specific latent variables induce disentangled reward representations, while sparsity over global latent factors acts as an implicit debiasing mechanism that suppresses spurious correlations. Together, this disentanglement-then-debiasing structure enables robust uncertainty-aware reward learning. To scale BNRM to modern LLMs, we develop an amortized variational inference network conditioned on deep model representations, allowing efficient end-to-end training. Extensive empirical results demonstrate that BNRM substantially mitigates reward over-optimization, improves robustness under distribution shifts, and yields more interpretable reward decompositions than strong baselines.
Read the full abstract ▾
All authors Zhibin Duan, Guowei Rong, Zhuo Li, Bo Chen, Mingyuan Zhou, Dandan Guo
Machine Learning
Pupillometry and Brain Dynamics for Cognitive Load in Working Memory
Nusaibah Farrukh, Malavika Pradeep, Akshay Sasi et al. (+2 more)
Cognitive load, the mental effort required during working
memory, is central to neuroscience, psychology, and human-computer interaction. Accurate assessment is vital for adaptive learning, clinical monitoring, and
brain-computer interfaces. Physiological signals such as pupillometry and electroencephalography are established biomarkers of cognitive load, but their comparative utility and practical integration as lightweight, wearable monitoring solutions remain underexplored.
EEG provides high temporal resolution of neural activity. Although non-invasive, it is technologically demanding and limited in wearability and cost due to its resource-intensive nature, whereas pupillometry is non-invasive, portable, and scalable. Existing studies often rely on deep learning models with limited interpretability and substantial computational expense. This study integrates feature-based and model-driven approaches to advance time-series analysis. Using the OpenNeuro 'Digit Span Task'
dataset, this study investigates cognitive load classification from
EEG and pupillometry. Feature-based approaches using Catch-22 features and classical machine learning models outperform deep learning in both binary and multiclass tasks. The findings demonstrate that pupillometry alone can compete with
EEG, serving as a portable and practical proxy for real-world applications. These results challenge the assumption that
EEG is necessary for load detection, showing that pupil dynamics combined with interpretable models and SHAP based feature analysis provide physiologically meaningful insights. This work supports the development of wearable, affordable cognitive monitoring systems for neuropsychiatry, education, and healthcare.
Read the full abstract ▾
All authors Nusaibah Farrukh, Malavika Pradeep, Akshay Sasi, Rahul Venugopal, Elizabeth Sherly
Machine Learning · Cross-listed from stat.ML
Highly Adaptive Principal Component Regression
Mingxun Wang, Alejandro Schuler, Mark van der Laan et al. (+1 more)
The Highly Adaptive Lasso (HAL) is a nonparametric regression method that achieves almost dimension-free convergence rates under minimal smoothness assumptions, but its implementation can be computationally prohibitive in high dimensions due to the large design
matrix it requires. The Highly Adaptive Ridge (HAR) has been proposed as a related ridge-regularized analogue. Building on both procedures, we introduce the Principal Component Highly Adaptive Lasso (PCHAL) and Principal Component Highly Adaptive Ridge (PCHAR). These
estimators use an outcome-blind principal-component reduction of the HAL basis, offering substantial computational gains over HAL while achieving empirical performance comparable to HAL and HAR. We also describe an early-stopped gradient descent variant, which provides a convenient form of smooth spectral regularization without explicitly selecting a hard principal-component cutoff. Finally, we uncover that under special circumstances, the HAL kernel is identical to the covariance function of Brownian motion.
Read the full abstract ▾
All authors Mingxun Wang, Alejandro Schuler, Mark van der Laan, Carlos García Meixide
Machine Learning
On the Role of Consistency Between Physics and Data in Physics-Informed Neural Networks
Nicolás Becerra-Zuniga, Lucas Lacasa, Eusebio Valero et al. (+1 more)
Physics-informed neural networks (PINNs) have gained significant
attention as a surrogate modeling strategy for partial differential equations (PDEs), particularly in regimes where labeled data are scarce and physical constraints can be leveraged to regularize the learning process. In practice, however, PINNs are frequently trained using experimental or numerical data that are not fully consistent with the governing equations due to measurement noise, discretization errors, or modeling assumptions. The implications of such data-to-PDE inconsistencies on the accuracy and convergence of PINNs remain insufficiently understood. In this work, we systematically analyze how data inconsistency fundamentally limits the attainable accuracy of PINNs. We introduce the concept of a consistency barrier, defined as an intrinsic lower bound on the error that arises from mismatches between the fidelity of the data and the exact enforcement of the PDE residual. To isolate and quantify this effect, we consider the 1D viscous Burgers equation with a manufactured analytical solution, which enables full control over data fidelity and residual errors. PINNs are trained using
datasets of progressively increasing numerical accuracy, as well as perfectly consistent analytical data. Results show that while the inclusion of the PDE residual allows PINNs to partially mitigate low-fidelity data and recover the dominant physical structure, the training process ultimately saturates at an error level dictated by the data inconsistency. When high-fidelity numerical data are employed, PINN solutions become indistinguishable from those trained on analytical data, indicating that the consistency barrier is effectively removed. These findings clarify the interplay between data quality and physics enforcement in PINNs providing practical guidance for the construction and interpretation of physics-informed surrogate models.
Read the full abstract ▾
All authors Nicolás Becerra-Zuniga, Lucas Lacasa, Eusebio Valero, Gonzalo Rubio
Machine Learning · Cross-listed from stat.ML
Bayesian Inference of Contextual Bandit Policies via Empirical Likelihood
Jiangrong Ouyang, Mingming Gong and Howard Bondell
Policy inference plays an essential role in the contextual bandit problem. In this paper, we use empirical likelihood to develop a
Bayesian inference method for the joint analysis of multiple contextual bandit policies in finite sample regimes. The proposed inference method is robust to small sample sizes and is able to provide accurate uncertainty measurements for policy value evaluation. In addition, it allows for flexible inferences on policy comparison with full uncertainty quantification. We demonstrate the effectiveness of the proposed inference method using Monte Carlo simulations and its application to an adolescent body mass index data set.
Read the full abstract ▾
Machine Learning
Hierarchical Zero-Order Optimization for Deep Neural Networks
Sansheng Cao, Zhengyu Ma and Yonghong Tian
Zeroth-order (ZO) optimization has long been favored for its biological plausibility and its capacity to handle non-differentiable objectives, yet its computational complexity has historically limited its application in deep neural networks. Challenging the conventional paradigm that gradients propagate layer-by-layer, we propose Hierarchical Zeroth-Order (HZO) optimization, a novel divide-and-conquer strategy that decomposes the depth dimension of the network. We prove that HZO reduces the query complexity from $O(ML^2)$ to $O(ML \log L)$ for a network of width $M$ and depth $L$, representing a significant leap over existing ZO methodologies. Furthermore, we provide a detailed error analysis showing that HZO maintains numerical stability by operating near the
unitary limit ($L_{lip} \approx 1$). Extensive evaluations on CIFAR-10 and ImageNet demonstrate that HZO achieves competitive accuracy compared to backpropagation.
Read the full abstract ▾
Machine Learning
dnaHNet: A Scalable and Hierarchical Foundation Model for Genomic Sequence Learning
Arnav Shah, Junzhe Li, Parsa Idehpour et al. (+7 more)
Genomic foundation models have the potential to decode DNA syntax, yet face a fundamental tradeoff in their input representation. Standard fixed-vocabulary tokenizers fragment biologically meaningful motifs such as codons and regulatory elements, while nucleotide-level models preserve biological coherence but incur prohibitive computational costs for long contexts. We introduce dnaHNet, a state-of-the-art tokenizer-free autoregressive model that segments and models genomic sequences end-to-end. Using a differentiable dynamic chunking mechanism, dnaHNet compresses raw nucleotides into latent
tokens adaptively, balancing compression with predictive accuracy. Pretrained on prokaryotic genomes, dnaHNet outperforms leading architectures including StripedHyena2 in scaling and efficiency. This recursive chunking yields quadratic FLOP reductions, enabling $>3 \times$ inference speedup over
Transformers. On zero-shot tasks, dnaHNet achieves superior performance in predicting
protein variant fitness and gene essentiality, while automatically discovering hierarchical biological structures without supervision. These results establish dnaHNet as a scalable, interpretable framework for next-generation genomic modeling.
Read the full abstract ▾
All authors Arnav Shah, Junzhe Li, Parsa Idehpour, Adibvafa Fallahpour, Brandon Wang, Sukjun Hwang, Bo Wang, Patrick D. Hsu, Hani Goodarzi, Albert Gu
Machine Learning
Learning Mixture Density via Natural Gradient Expectation Maximization
Yutao Chen, Jasmine Bayrooti and Steven Morad
Mixture density networks are neural networks that produce Gaussian mixtures to represent continuous
multimodal conditional densities. Standard training procedures involve maximum likelihood estimation using the negative log-likelihood (NLL) objective, which suffers from slow convergence and mode collapse. In this work, we improve the optimization of mixture density networks by integrating their information geometry. Specifically, we interpret mixture density networks as deep latent-variable models and analyze them through an expectation maximization framework, which reveals surprising theoretical connections to natural gradient descent. We then exploit such connections to derive the natural gradient expectation maximization (nGEM) objective. We show that empirically nGEM achieves up to 10$\times$ faster convergence while adding almost zerocomputational overhead, and scales well to high-dimensional data where NLL otherwise fails.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.AI
Neuro-symbolic Action Masking for Deep Reinforcement Learning
Shuai Han, Mehdi Dastani and Shihan Wang
Deep reinforcement learning (DRL) may explore infeasible actions during training and execution. Existing approaches assume a symbol grounding function that maps high-dimensional states to consistent symbolic representations and a manually specified action masking techniques to constrain actions. In this paper, we propose Neuro-symbolic Action Masking (NSAM), a novel framework that automatically learn symbolic models, which are consistent with given domain constraints of high-dimensional states, in a minimally supervised manner during the DRL process. Based on the learned symbolic model of states, NSAM learns action masks that rules out infeasible actions. NSAM enables end-to-end integration of symbolic
reasoning and deep policy optimization, where improvements in symbolic grounding and policy learning mutually reinforce each other. We evaluate NSAM on multiple domains with constraints, and experimental results demonstrate that NSAM significantly improves sample efficiency of DRL
agent while substantially reducing constraint violations.
Read the full abstract ▾
Machine Learning
Roughness-Informed Federated Learning
Mohammad Partohaghighi, Roummel Marcia, Bruce J. West et al. (+1 more)
Federated Learning (FL) enables collaborative model training across distributed clients while preserving data
privacy, yet faces challenges in non-independent and identically distributed (non-IID) settings due to client drift, which impairs convergence. We propose RI-FedAvg, a novel FL
algorithm that mitigates client drift by incorporating a Roughness Index (RI)-based regularization term into the local objective, adaptively penalizing updates based on the fluctuations of local loss landscapes. This paper introduces RI-FedAvg, leveraging the RI to quantify the roughness of high-dimensional loss functions, ensuring robust optimization in heterogeneous settings. We provide a rigorous convergence analysis for non-convex objectives, establishing that RI-FedAvg converges to a stationary point under standard assumptions. Extensive experiments on MNIST, CIFAR-10, and CIFAR-100 demonstrate that RI-FedAvg outperforms state-of-the-art baselines, including FedAvg, FedProx, FedDyn, SCAFFOLD, and DP-FedAvg, achieving higher accuracy and faster convergence in non-IID scenarios. Our results highlight RI-FedAvg's potential to enhance the robustness and efficiency of
federated learning in practical, heterogeneous environments.
Read the full abstract ▾
All authors Mohammad Partohaghighi, Roummel Marcia, Bruce J. West, YangQuan Chen
Machine Learning · Cross-listed from cs.RO
Flow-Enabled Generalization to Human Demonstrations in Few-Shot Imitation Learning
Runze Tang and Penny Sweetser
Imitation Learning (IL) enables
robots to learn complex skills from demonstrations without explicit task modeling, but it typically requires large amounts of demonstrations, creating significant collection costs. Prior work has investigated using flow as an intermediate representation to enable the use of human
videos as a substitute, thereby reducing the amount of required
robot demonstrations. However, most prior work has focused on the flow, either on the object or on specific points of the
robot/hand, which cannot describe the motion of interaction. Meanwhile, relying on flow to achieve generalization to scenarios observed only in human
videos remains limited, as flow alone cannot capture precise motion details. Furthermore, conditioning on scene observation to produce precise actions may cause the flow-conditioned policy to overfit to training tasks and weaken the generalization indicated by the flow. To address these gaps, we propose SFCrP, which includes a Scene Flow prediction model for Cross-embodiment learning (SFCr) and a Flow and Cropped point cloud conditioned Policy (FCrP). SFCr learns from both
robot and human
videos and predicts any point trajectories. FCrP follows the general flow motion and adjusts the action based on observations for precision tasks. Our method outperforms SOTA baselines across various real-world task settings, while also exhibiting strong spatial and instance generalization to scenarios seen only in human
videos.
Read the full abstract ▾
Machine Learning
TRACE: Theoretical Risk Attribution under Covariate-shift Effects
Hosein Anjidani, S. Yahya S. R. Tehrani, Mohammad Mahdi Mojahedian et al. (+1 more)
When a source-trained model $Q$ is replaced by a model $\tilde{Q}$ trained on shifted data, its performance on the source domain can change unpredictably. To address this, we study the two-model risk change, $\Delta R := R_P(\tilde{Q}) - R_P(Q)$, under
covariate shift. We introduce TRACE (Theoretical Risk Attribution under
Covariate-shift Effects), a framework that decomposes $|\Delta R|$ into an interpretable upper bound. This decomposition disentangles the risk change into four actionable factors: two generalization gaps, a model change penalty, and a
covariate shift penalty, transforming the bound into a powerful diagnostic tool for understanding why performance has changed. To make TRACE a fully computable diagnostic, we instantiate each term. The
covariate shift penalty is estimated via a model sensitivity factor (from high-
quantile input gradients) and a data-shift measure; we use feature-space Optimal Transport (OT) by default and provide a robust alternative using Maximum Mean Discrepancy (MMD). The model change penalty is controlled by the average output distance between the two models on the target sample. Generalization gaps are estimated on held-out data. We validate our framework in an idealized linear regression setting, showing the TRACE bound correctly captures the scaling of the true risk difference with the magnitude of the shift. Across synthetic and
vision benchmarks, TRACE diagnostics are valid and maintain a strong monotonic relationship with the true performance degradation. Crucially, we derive a deployment gate score that correlates strongly with $|\Delta R|$ and achieves high AUROC/AUPRC for gating decisions, enabling safe, label-efficient model replacement.
Read the full abstract ▾
All authors Hosein Anjidani, S. Yahya S. R. Tehrani, Mohammad Mahdi Mojahedian, Mohammad Hossein Yassaee
Machine Learning · Cross-listed from stat.ML
Deep Bootstrap
Jinyuan Chang, Yuling Jiao, Lican Kang et al. (+1 more)
In this work, we propose a novel deep bootstrap framework for nonparametric regression based on conditional
diffusion models. Specifically, we construct a conditional
diffusion model to learn the distribution of the response variable given the
covariates. This model is then used to generate bootstrap samples by pairing the original
covariates with newly synthesized responses. We reformulate nonparametric regression as conditional sample mean estimation, which is implemented directly via the learned conditional
diffusion model. Unlike traditional bootstrap methods that decouple the estimation of the conditional distribution, sampling, and nonparametric regression, our approach integrates these components into a unified generative framework. With the expressive capacity of
diffusion models, our method facilitates both efficient sampling from high-dimensional or
multimodal distributions and accurate nonparametric estimation. We establish rigorous theoretical guarantees for the proposed method. In particular, we derive optimal end-to-end convergence rates in the Wasserstein distance between the learned and target conditional distributions. Building on this foundation, we further establish the convergence guarantees of the resulting bootstrap procedure. Numerical studies demonstrate the effectiveness and scalability of our approach for complex regression tasks.
Read the full abstract ▾
All authors Jinyuan Chang, Yuling Jiao, Lican Kang, Junjie Shi
Machine Learning
Neural Additive Experts: Context-Gated Experts for Controllable Model Additivity
Guangzhi Xiong, Sanchit Sinha and Aidong Zhang
The trade-off between interpretability and accuracy remains a core challenge in machine learning. Standard Generalized Additive Models (GAMs) offer clear feature attributions but are often constrained by their strictly additive nature, which can limit predictive performance. Introducing feature interactions can boost accuracy yet may obscure individual feature contributions. To address these issues, we propose Neural Additive Experts (NAEs), a novel framework that seamlessly balances interpretability and accuracy. NAEs employ a mixture of experts framework, learning multiple specialized networks per feature, while a dynamic gating mechanism integrates information across features, thereby relaxing rigid additive constraints. Furthermore, we propose targeted regularization techniques to mitigate variance among expert predictions, facilitating a smooth transition from an exclusively additive model to one that captures intricate feature interactions while maintaining clarity in feature attributions. Our theoretical analysis and experiments on synthetic data illustrate the model's flexibility, and extensive evaluations on real-world
datasets confirm that NAEs achieve an optimal balance between predictive accuracy and transparent, feature-level explanations. The code is available at https://github.com/Teddy-XiongGZ/NAE.
Read the full abstract ▾
Machine Learning
When Gradient Clipping Becomes a Control Mechanism for Differential Privacy in Deep Learning
Mohammad Partohaghighi, Roummel Marcia, Bruce J. West et al. (+1 more)
Privacy-preserving training on sensitive data commonly relies on differentially private
stochastic optimization with gradient clipping and Gaussian noise. The clipping threshold is a critical control knob: if set too small, systematic over-clipping induces optimization bias; if too large, injected noise dominates updates and degrades accuracy. Existing adaptive clipping methods often depend on per-example gradient norm statistics, adding computational overhead and introducing sensitivity to
datasets and architectures. We propose a control-driven clipping strategy that adapts the threshold using a lightweight, weight-only spectral diagnostic computed from model parameters. At periodic probe steps, the method analyzes a designated weight
matrix via spectral decomposition and estimates a heavy-tailed spectral indicator associated with training stability. This indicator is smoothed over time and fed into a bounded feedback controller that updates the clipping threshold multiplicatively in the log domain. Because the controller uses only parameters produced during
privacy-preserving training, the resulting threshold updates are post-processing and do not increase
privacy loss beyond that of the underlying DP optimizer under standard composition accounting.
Read the full abstract ▾
All authors Mohammad Partohaghighi, Roummel Marcia, Bruce J. West, YangQuan Chen
Machine Learning
LLM-Based Scientific Equation Discovery via Physics-Informed Token-Regularized Policy Optimization
Boxiao Wang, Kai Li, Tianyi Liu et al. (+4 more)
Symbolic regression aims to distill mathematical equations from observational data. Recent approaches have successfully leveraged Large
Language Models (LLMs) to generate equation hypotheses, capitalizing on their vast pre-trained scientific priors. However, existing frameworks predominantly treat the LLM as a static generator, relying on
prompt-level guidance to steer exploration. This paradigm fails to update the model's internal representations based on search feedback, often yielding physically inconsistent or mathematically redundant expressions. In this work, we propose PiT-PO (Physics-informed
Token-regularized Policy Optimization), a unified framework that evolves the LLM into an adaptive generator via reinforcement learning. Central to PiT-PO is a dual-constraint mechanism that rigorously enforces hierarchical physical validity while simultaneously applying fine-grained,
token-level penalties to suppress redundant structures. Consequently, PiT-PO aligns LLM to produce equations that are both scientifically consistent and structurally parsimonious. Empirically, PiT-PO achieves state-of-the-art performance on standard
benchmarks and successfully discovers novel
turbulence models for challenging fluid dynamics problems. We also demonstrate that PiT-PO empowers small-scale models to outperform closed-source
giants, democratizing access to high-performance scientific discovery.
Read the full abstract ▾
All authors Boxiao Wang, Kai Li, Tianyi Liu, Chen Li, Junzhe Wang, Yifan Zhang, Jian Cheng
Machine Learning
Gauss-Newton Unlearning for the LLM Era
Lev McKinney, Anvith Thudi, Juhan Bae et al. (+4 more)
Standard large
language model training can create models that produce outputs their trainer deems unacceptable in deployment. The probability of these outputs can be reduced using methods such as LLM unlearning. However, unlearning a set of data (called the forget set) can degrade model performance on other distributions where the trainer wants to retain the model's behavior. To improve this trade-off, we demonstrate that using the forget set to compute only a few uphill Gauss-Newton steps provides a conceptually simple, state-of-the-art unlearning approach for LLMs. While Gauss-Newton steps adapt Newton's method to non-linear models, it is non-trivial to efficiently and accurately compute such steps for LLMs. Hence, our approach crucially relies on parametric Hessian approximations such as Kronecker-Factored Approximate
Curvature (K-FAC). We call this combined approach K-FADE (K-FAC for Distribution Erasure). Our evaluation on the WMDP and ToFU
benchmarks demonstrates that K-FADE suppresses outputs from the forget set and approximates, in output space, the results of retraining without the forget set. Critically, our method does this while altering the outputs on the retain set less than previous methods. This is because K-FADE transforms a constraint on the model's outputs across the entire retain set into a constraint on the model's weights, allowing the
algorithm to minimally change the model's behavior on the retain set at each step. Moreover, the unlearning updates computed by K-FADE can be reapplied later if the model undergoes further training, allowing unlearning to be cheaply maintained.
Read the full abstract ▾
All authors Lev McKinney, Anvith Thudi, Juhan Bae, Tara Rezaei, Nicolas Papernot, Sheila A. McIlraith, Roger Grosse
Machine Learning
Online Min-Max Optimization: From Individual Regrets to Cumulative Saddle Points
Abhijeet Vyas and Brian Bullins
We propose and study an online version of min-max optimization based on cumulative saddle points under a variety of performance measures beyond convex-concave settings. After first observing the incompatibility of (static) Nash
equilibrium (SNE-Reg$_T$) with individual regrets even for strongly convex-strongly concave functions, we propose an alternate static duality gap (SDual-Gap$_T$) inspired by the online convex optimization (OCO) framework. We provide
algorithms that, using a reduction to classic OCO problems, achieve bounds for SDual-Gap$_T$~and a novel dynamic saddle point regret (DSP-Reg$_T$), which we suggest naturally represents a min-max version of the dynamic regret in OCO. We derive our bounds for SDual-Gap$_T$~and DSP-Reg$_T$~under strong convexity-strong concavity and a min-max notion of exponential concavity (min-max EC), and in addition we establish a class of functions satisfying min-max EC~that captures a two-player variant of the classic portfolio selection problem. Finally, for a dynamic notion of regret compatible with individual regrets, we derive bounds under a two-sided Polyak-Łojasiewicz (PL) condition.
Read the full abstract ▾
Machine Learning
Contrastive Learning for Multi Label ECG Classification with Jaccard Score Based Sigmoid Loss
Junichiro Takahashi, Masataka Sato, Satoshi Kodeta et al. (+1 more)
Recent advances in large
language models (LLMs) have enabled the development of
multimodal medical AI. While models such as MedGemini achieve high accuracy on VQA tasks like USMLE MM, their performance on ECG based tasks remains limited, and some models, such as MedGemma, do not support ECG data at all. Interpreting ECGs is inherently challenging, and diagnostic accuracy can vary depending on the interpreter's experience. Although echocardiography provides rich diagnostic information, it requires specialized equipment and personnel, limiting its availability. In this study, we focus on constructing a robust ECG encoder for
multimodal pretraining using real world hospital data. We employ SigLIP, a CLIP based model with a sigmoid based loss function enabling multi label prediction, and introduce a modified loss function tailored to the multi label nature of ECG data. Experiments demonstrate that incorporating medical knowledge in the
language model and applying the modified loss significantly improve multi label ECG classification. To further enhance performance, we increase the
embedding dimensionality and apply random cropping to mitigate data drift. Finally, per label analysis reveals which ECG findings are easier or harder to predict. Our study provides a foundational framework for developing medical models that utilize ECG data.
Read the full abstract ▾
All authors Junichiro Takahashi, Masataka Sato, Satoshi Kodeta, Norihiko Takeda
Machine Learning
$\mu$pscaling small models: Principled warm starts and hyperparameter transfer
Yuxin Ma, Nan Chen, Mateo Díaz et al. (+3 more)
Modern large-scale neural networks are often trained and released in multiple sizes to accommodate diverse inference budgets. To improve efficiency, recent work has explored model upscaling: initializing larger models from trained smaller ones to accelerate convergence. However, this method can be sensitive to hyperparameters that need to be tuned at the target upscaled model size, which is prohibitively costly to do directly. It remains unclear whether tuning hyperparameters on smaller models and extrapolating via scaling laws is sound in this setting. We address this with principled approaches to width-based upscaling and efficient hyperparameter tuning in this setting. Motivated by $\mu$P and any-dimensional architectures, we introduce a general upscaling method that, like Net2Net, copies and perturbs weights, but uses theoretically grounded, width-dependent scalings for the perturbation noise and optimizer hyperparameters. First, we prove that under zero perturbation, the upscaled model is functionally equivalent to the base model throughout training. Second, we extend the $\mu$P theory to enable infinite-width limit analysis and establish hyperparameter transfer for upscaled models, greatly reducing the tuning cost. We empirically demonstrate that this method is effective on realistic
datasets and architectures.
Read the full abstract ▾
All authors Yuxin Ma, Nan Chen, Mateo Díaz, Soufiane Hayou, Dmitriy Kunisky, Soledad Villar
Machine Learning
Bridging the Compression-Precision Paradox: A Hybrid Architecture for Clinical EEG Report Generation with Guaranteed Measurement Accuracy
Wuyang Zhang, Zhen Luo, Chuqiao Gu et al. (+4 more)
Automated
EEG monitoring requires clinician-level precision for seizure detection and reporting. Clinical
EEG recordings exceed LLM context windows, requiring extreme compression (400:1+ ratios) that destroys fine-grained temporal precision. A 0.5 Hz error distinguishes absence epilepsy from Lennox-Gastaut syndrome. LLMs lack inherent time-series comprehension and rely on statistical associations from compressed representations. This dual limitation causes systems to hallucinate clinically incorrect measurement values. We separate measurement extraction from text generation. Our hybrid architecture computes exact clinical values via signal processing before compression, employs a cross-modal bridge for
EEG-to-
language translation, and uses parameter-efficient fine-tuning with constrained decoding around frozen slots. Multirate sampling maintains long-range context while preserving event-level precision. Evaluation on TUH and CHB-MIT
datasets achieves 60% fewer false alarms, 50% faster detection, and sub-clinical measurement precision. This is the first system guaranteeing clinical measurement accuracy in automated
EEG reports.
Read the full abstract ▾
All authors Wuyang Zhang, Zhen Luo, Chuqiao Gu, Jianming Ma, Yebo Cao, Wangming Yuan, Yinzhi Jin
Machine Learning · Cross-listed from cs.IT
Predictive-State Communication: Innovation Coding and Reconciliation under Delay
Ozgur Ercetin and Mohaned Chraiti
Shannon theory models communication as the reliable transfer of symbol sequences, with performance governed by capacity and rate-distortion limits. When both endpoints possess strong predictors – as in modern large
language models and related generative priors – literal symbol transport is no longer the only operational regime. We propose predictive-state communication (PSC), in which the transmitter and receiver maintain an explicit shared predictive state, and the physical channel is used primarily to convey innovations, i.e., corrective information that reconciles the receiver's provisional trajectory with the transmitter's realized trajectory. This viewpoint replaces
entropy-rate accounting by cross-
entropy accounting under model mismatch, and it introduces feasibility constraints that depend jointly on capacity, delay, and perceptual continuity requirements; the resulting operating set is typically a bounded perception-capacity band rather than a one-sided threshold. We outline the protocol and architectural implications (state identifiers, anchors, bounded rollback, and patch-based updates) and provide a stylized illustrative example to visualize the induced feasibility region and its dependence on predictive quality.
Read the full abstract ▾
Machine Learning · Cross-listed from math.NA
FastLSQ: Solving PDEs in One Shot via Fourier Features with Exact Analytical Derivatives
Antonin Sulc
We present FastLSQ, a framework for PDE solving and inverse problems built on trigonometric random
Fourier features with exact analytical derivatives. Trigonometric features admit closed-form derivatives of any order in $\mathcal{O}(1)$, enabling graph-free operator assembly without autodiff. Linear PDEs: one least-squares call; nonlinear: Newton–Raphson reusing analytical assembly. On 17 PDEs (1–6D), FastLSQ achieves $10^{-7}$ in 0.07s (linear) and $10^{-8}$–$10^{-9}$ in $<$9s (nonlinear), orders of magnitude faster and more accurate than iterative PINNs. Analytical higher-order derivatives yield a differentiable digital twin; we demonstrate inverse problems (heat-source, coil recovery) and PDE discovery. Code: github.com/sulcantonin/FastLSQ and pip install fastlsq.
Read the full abstract ▾
Machine Learning
What Makes Value Learning Efficient in Residual Reinforcement Learning?
Guozheng Ma, Lu Li, Haoyu Wang et al. (+3 more)
Residual reinforcement learning (RL) enables stable online refinement of expressive pretrained policies by freezing the base and learning only bounded corrections. However, value learning in residual RL poses unique challenges that remain poorly understood. In this work, we identify two key bottlenecks: cold start pathology, where the critic lacks knowledge of the value landscape around the base policy, and structural scale mismatch, where the residual contribution is dwarfed by the base action. Through systematic investigation, we uncover the mechanisms underlying these bottlenecks, revealing that simple yet principled solutions suffice: base-policy transitions serve as an essential value anchor for implicit warmup, and critic normalization effectively restores representation sensitivity for discerning value differences. Based on these insights, we propose DAWN (Data-Anchored Warmup and Normalization), a minimal approach targeting efficient value learning in residual RL. By addressing these bottlenecks, DAWN demonstrates substantial efficiency gains across diverse
benchmarks, policy architectures, and observation modalities.
Read the full abstract ▾
All authors Guozheng Ma, Lu Li, Haoyu Wang, Zixuan Liu, Pierre-Luc Bacon, Dacheng Tao
Machine Learning · Cross-listed from stat.ML
Why Agentic Theorem Prover Works: A Statistical Provability Theory of Mathematical Reasoning Models
Sho Sonoda, Shunta Akiyama and Yuya Uezato
Agentic
theorem provers combine a
reasoning model,
retrieval, search, and a proof assistant verifier, yet it remains unclear which components actually improve finite-budget proof success and why they help on real mathematical workloads. We study this question through statistical provability: the probability of reaching a verified proof within a budget on a specified stream of
theorem instances. We model formal proof search as a finite-horizon reachability MDP with deterministic verifier dynamics, and show that under a faithful state abstraction the optimal success probability coincides with ordinary syntactic provability. We then analyze a simple but practically important pipeline: depth-wise offline action-value regression followed by greedy test-time proving. Our main
theorem bounds the provability gap between the learned prover and the optimal prover by an occupancy-weighted sum of uniform action-value errors; in the common uniform-error reading, the leading complexity multiplier is the learned prover's average truncated proof length. The error decomposes into approximation error,
geometric coverage of the training distribution, and Monte Carlo label noise, and improves to a fast rate under an action-gap margin condition. The result gives a component-sensitive account of why verifier feedback,
retrieval, representation geometry, and proof-shortening mechanisms help on biased
theorem workloads, without contradicting classical worst-case hardness.
Read the full abstract ▾
Machine Learning · Cross-listed from stat.ML
Statistical Inference and Learning for Shapley Additive Explanations (SHAP)
Justin Whitehouse, Ayush Sawarni and Vasilis Syrgkanis
The SHAP (short for Shapley additive explanation) framework has become an essential tool for attributing importance to variables in predictive tasks. In model-agnostic settings, SHAP uses the concept of Shapley values from cooperative game theory to fairly allocate credit to the features in a vector $X$ based on their contribution to an outcome $Y$. While the explanations offered by SHAP are local by nature, learners often need global measures of feature importance in order to improve model explainability and perform feature selection. The most common approach for converting these local explanations into global ones is to compute either the mean absolute SHAP or mean squared SHAP. However, despite their ubiquity, there do not exist approaches for performing statistical inference on these quantities. In this paper, we take a semi-parametric approach for calibrating confidence in estimates of the $p$th powers of Shapley additive explanations. We show that, by treating the SHAP curve as a nuisance function that must be estimated from data, one can reliably construct asymptotically normal estimates of the $p$th powers of SHAP. When $p \geq 2$, we show a de-biased
estimator that combines U-statistics with Neyman orthogonal scores for functionals of nested regressions is asymptotically normal. When $1 \leq p < 2$ (and the hence target parameter is not twice differentiable), we construct de-biased U-statistics for a smoothed alternative. In particular, we show how to carefully tune the temperature parameter of the smoothing function in order to obtain inference for the true, unsmoothed $p$th power. We complement these results by presenting a Neyman orthogonal loss that can be used to learn the SHAP curve via empirical risk minimization and discussing excess risk guarantees for commonly used function classes.
Read the full abstract ▾
Machine Learning · Cross-listed from stat.ML
From Collapse to Improvement: Statistical Perspectives on the Evolutionary Dynamics of Iterative Training on Contaminated Sources
Soham Bakshi and Sunrit Chakraborty
The problem of model collapse has presented new challenges in iterative training of generative models, where such training with synthetic data leads to an overall degradation of performance. This paper looks at the problem from a statistical viewpoint, illustrating that one can actually hope for improvement when models are trained on data contaminated with synthetic samples, as long as there is some amount of fresh information from the true target distribution. In particular, we consider iterative training on samples sourced from a mixture of the true target and synthetic distributions. We analyze the entire iterative
evolution in a next-
token prediction
language model, capturing how the interplay between the mixture weights and the sample size controls the overall long-term performance. With non-trivial mixture weight of the true distribution, even if it decays over time, simply training the model in a contamination-agnostic manner with appropriate sample sizes can avoid collapse and even recover the true target distribution under certain conditions. Simulation studies support our findings and also show that such behavior is more general for other classes of models.
Read the full abstract ▾