cs.LG · 2026-05-27 · No. 10

Machine Learning, 2026-05-27.

66 new papers in cs.LG. Titles, authors, abstracts. Links to arXiv. Want this in your inbox every morning? Subscribe →

01 — The papers

66 entries
  1. 01

    MobileMoE: Scaling On-Device Mixture of Experts

    Yanbei Chen, Hanxian Huang, Ernie Chang, Jacob Szwejbka, Digant Desai, Zechun Liu, Vikas Chandra, Raghuraman Krishnamoorthi

    cs.LG · cs.AI · cs.CL

    Mixture-of-Experts (MoE) has become the de facto architecture for hundred-billion-parameter language models, yet its advantages at sub-billion scales for on-device deployment remain largely unexplored. To close this gap, we present MobileMoE, a family of on-device MoE language models with sub-billion active parameters (0.3-0.9B active and 1.3-5.3B total) that establish a new Pareto frontier for on-device LLMs. We first formulate an on-device...

    arxiv.org/abs/2605.27358 · PDF

  2. 02

    Guiding LLM Post-training Data Engineering with Model Internals from Sparse Autoencoders

    Yi Jing, Zao Dai, Jinwu Hu, Zijun Yao, Lei Hou, Juanzi Li, Xiaozhi Wang

    cs.LG · cs.AI · cs.CL

    Model internals encode rich information about how a large language model (LLM) processes its training data; however, post-training data engineering largely relies on external signals and ignores rich intrinsic signals lying in model internals. We propose SAERL, a data engineering framework for LLM reinforcement learning (RL). It models three intrinsic data properties: diversity, difficulty, and quality, using model internals extracted with...

    arxiv.org/abs/2605.27354 · PDF

  3. 03

    From Scores to Gibbs Correctors: Accelerating Uniform-Rate Discrete Diffusion Models

    Yuchen Liang, Ness Shroff, Yingbin Liang

    cs.LG · stat.ML

    Discrete diffusion models have achieved strong empirical performance in text and other symbolic domains, but, especially for uniform-rate models, they often require many steps to generate a single sample. Existing acceleration methods either rely on training additional quantities or suffer from slow mixing. In this work, we propose a novel Gibbs-based corrector for discrete diffusion models, termed Gibbs-Accelerated Discrete Diffusion (GADD)....

    arxiv.org/abs/2605.27352 · PDF

  4. 04

    Probabilistic Smoothing with Ratio-Monotone Transforms for Global Optimization

    Kukyoung Jang, Taehyun Cho, Junrui Zhang, Ping Xu, Kyungjae Lee

    cs.LG · math.OC

    Probabilistic smoothing is a standard tool for global optimization, but existing methods rely on Gaussian kernels and specific transforms, often resulting in strong hyperparameter sensitivity and limited robustness. We propose a general smoothing framework that combines flexible symmetric unimodal kernels with monotonic ratio-based transformations. Under mild conditions, we show that the smoothed objective preserves the global maximizer and...

    arxiv.org/abs/2605.27316 · PDF

  5. 05

    Greening AI Inference with Accuracy and Latency-aware User Incentives

    Vasilios A. Siris, Adamantia Stamou, George D. Stamoulis, Konstantinos Varsos, Ramin Khalili

    cs.LG · cs.OH

    The widespread use of AI services has raised concerns for its environmental sustainability, towards which recent studies have identified carbon emissions of AI inference as the major contributor. This paper introduces a framework for designing AI inference incentives based on the users' valuation for inference quality and latency, together with their environmental consciousness, while accounting for the tradeoff between carbon emissions and...

    arxiv.org/abs/2605.27309 · PDF

  6. 06

    Normal Guidance is what Attention Needs

    Ethan Harvey, Dennis Johan Loevlie, Michael C. Hughes

    cs.LG

    We consider training classifiers for 3D medical images using only one binary label for the entire volume rather than a label for each 2D slice. In such weakly supervised settings, can we learn accurate classifiers for slice-level predictions? Attention-based multiple instance learning (MIL) can produce an attention score for every slice. Yet recent work demonstrates that a simple center-focused baseline that ignores image content can...

    arxiv.org/abs/2605.27306 · PDF

  7. 07

    BASIS: Batchwise Advantage Estimation from Single-Rollout Information Sharing for LLM Reasoning

    Shijin Gong, Erhan Xu, Kai Ye, Francesco Quinzan, Giulia Livieri, Chengchun Shi

    cs.LG · stat.ML

    Reinforcement learning with verifiable rewards has become a standard recipe for improving the reasoning abilities of large language models. Existing algorithms face a tradeoff between computational efficiency and sample efficiency in value estimation and policy learning. We introduce BASIS, a critic-free post-training algorithm designed to address this tradeoff. At each online training step, BASIS samples only one rollout per prompt, but...

    arxiv.org/abs/2605.27293 · PDF

  8. 08

    Detectability in Diversity: Improved Canary Crafting for Privacy Auditing in One Run

    Mathieu Dagréou, Aurélien Bellet

    cs.LG · stat.ML

    Privacy auditing aims to empirically assess privacy leakage in machine learning models using membership inference attacks (MIAs), and to derive lower bounds on differential privacy (DP) parameters. Recent one-run auditing methods address the high cost of standard approaches by relying on a single training run with multiple "canary" points whose inclusion or exclusion must be detected by the auditor. In this work, we study the problem of...

    arxiv.org/abs/2605.27292 · PDF

  9. 09

    Falcon-X: A Time Series Foundation Model for Heterogeneous Multivariate Modeling

    Yiding Liu, Yifan Hu, Hongjie Xia, Peiyuan Liu, Hongzhou Chen, Xilin Dai, Zewei Dong, Jiang-Ming Yang

    cs.LG · cs.AI

    Time series foundation models (TSFMs) are transforming the forecasting paradigm through large-scale cross-domain pretraining. However, most existing TSFMs remain univariate, and recent efforts to enable cross-variate modeling still operate directly within the raw variate space. This design introduces fundamental limitations in semantic alignment and relational expressivity. Specifically, raw-space group mixing lacks a dedicated mechanism to...

    arxiv.org/abs/2605.27286 · PDF

  10. 10

    Causal Risk Minimization for High-Dimensional Treatments

    Nikita Dhawan, Arnav Paruthi, Andrew Kim, Lovedeep Gondara, Jekaterina Novikova, Chris J. Maddison

    cs.LG · stat.ML

    Predicting the effect of interventions with many possible variations, e.g., therapeutic content that affects mental health outcomes or an earnings call transcript that drives movement in share price, is useful across several domains. However, classical causal estimators tend to assume that all possible interventions are observed, which is infeasible when interventions vary widely, for instance, in the space of all text strings. We adapt a...

    arxiv.org/abs/2605.27281 · PDF

  11. 11

    Transfer Learning using 66 Diseases for Disease Forecasting Applications

    Lauren J Beesley, Alexander C Murph, Dave Osthus, Lauren A Castro

    cs.LG · stat.AP

    Disease forecasting models typically rely on a single data stream, making models brittle when histories are short or noisy. Recent top-performing models have shown that synthesizing multiple reporting systems for the same disease improves performance. Other recent work takes this idea a step further, using transfer learning to train a forecasting model for one disease using data from a different disease. We expand upon each of these...

    arxiv.org/abs/2605.27269 · PDF

  12. 12

    Kan Extension Transformers: A Categorical Unification of Attention, Diffusion, and Predict-Detach Self-Conditioning

    Sridhar Mahadevan

    cs.LG

    We propose Kan Extension Transformers (KETs) as a unifying categorical framework for a diverse group of Transformer implementations. The core claim is that a Transformer layer can be viewed as a weighted structured extension operator: standard attention is the singleton-neighborhood case, Geometric Transformer style incidence mixing is a sparse edge-restricted case, and KET is the higher-order simplicial case. This lens also clarifies a...

    arxiv.org/abs/2605.27259 · PDF

  13. 13

    LUCoS: Latent Unsupervised Context Selection for Tabular Foundation Models

    Oroel Ipas, Guillermo Gomez-Trenado, Rocío Romero-Zaliz, Isaac Triguero

    cs.LG · cs.AI

    Selecting which instances to label is a key challenge in low-label tabular learning. For recent Tabular Foundation Models such as TabPFN, context selection directly determines predictive performance. Supervised oracle experiments show that carefully chosen labeled context sets can strongly outperform random selection under the same labeling budget. However, the cold-start setting, where instances must be selected before any labels are...

    arxiv.org/abs/2605.27254 · PDF

  14. 14

    Symbolic Regression via Latent Iterative Refinement

    Xieting Chu, Sriram Vishwanath, Vijay Ganesh

    cs.LG

    Symbolic regression (SR) seeks closed-form mathematical expressions that fit observed data. Neural SR methods amortize the search by training an encoder to map observations directly to expressions in a single pass, but this amortized inference leaves a residual amortization gap between its one-shot prediction and the true posterior. We propose Latent Equation Embedding (LEE), a framework that closes this gap through iterative amortized...

    arxiv.org/abs/2605.27245 · PDF

  15. 15

    Explainable Comparison of Feature-Based and Deep Learning Models for TROPOMI Methane Plume Screening

    Solomiia Kurchaba, Joannes D. Maasakkers, Berend J. Schuit, Ilse Aben

    cs.LG · physics.ao-ph

    Continuous and global detection of large methane emissions is a crucial step for global warming mitigation. Satellite observations, such as from S5P/TROPOMI, combined with plume detection algorithms, can play a key role in this effort. However, not all TROPOMI plume detections that look like methane emission plumes are the result of actual emissions. A significant part of the plume-like features in the data are retrieval artifacts. Such...

    arxiv.org/abs/2605.27236 · PDF

  16. 16

    Nonlinear Data Integration via Kernel Methods for Data Collaboration Analysis

    Yamato Suetake, Yuta Kawakami, Shunnosuke Ikeda, Yuichi Takano

    cs.LG · stat.ML

    Collaborative analysis of decentralized confidential datasets is important, but direct sharing of original datasets is often restricted by privacy and institutional constraints. Data collaboration (DC) analysis transforms each dataset into privacy-preserving intermediate representations via party-specific obfuscation functions and integrates them into common collaboration representations using an anchor dataset. However, many existing DC...

    arxiv.org/abs/2605.27219 · PDF

  17. 17

    The Role of Causal Features in Strategic Classification for Robustness and Alignment

    Antonio Gois, Sophia Gunluk, Nir Rosenfeld, Nidhi Hegde, Simon Lacoste-Julien, Dhanya Sridhar

    cs.LG · stat.ML

    In strategic classification, an institution (e.g., a bank) anticipates adaptation from users who change their features to increase utility in a classification task (e.g., loan repayment). Since a key challenge is the distribution shift induced by users, we turn to causal models, which have been shown to bound the worst-case out-of-distribution (OOD) risk, and establish several new results that link causality and strategic classification....

    arxiv.org/abs/2605.27163 · PDF

  18. 18

    Deep-layer limit and stability analysis of the basic forward-backward-splitting induced network (II): learning problems

    Xuan Lin, Chunlin Wu

    cs.LG · cs.AI

    Deep unfolding neural networks derived from iterative optimization schemes and numerical ordinary/partial differential equations (ODEs/PDEs) have attracted much attention in data science over the last decade. Therein, numerous important network architectures were constructed from the basic forward-backward-splitting (FBS) algorithm. In this paper, we continue our research on the most basic FBS-induced network, an architecture unrolled from...

    arxiv.org/abs/2605.27133 · PDF

  19. 19

    DEI: Diversity in Evolutionary Inference for Quality-Diversity Search

    John Donaghy, Shikhar Rastogi

    cs.LG · cs.AI

    We present DEI: Diversity in Evolutionary Inference, a distributed Quality-Diversity (QD) search framework that assigns heterogeneous large language models (LLMs) as mutation operators across peer nodes communicating with non-blocking collective operations. Unlike homogeneous parallel search, which replicates a single model's inductive biases across all workers, DEI treats each LLM's distinct creative prior as a complementary source of...

    arxiv.org/abs/2605.27130 · PDF

  20. 20

    High-Quality Synthetic Financial Time-Series using a GAN-Diffusion Framework

    Giuseppe Masi, Andrea Coletta, Novella Bartolini

    cs.LG · cs.AI

    In recent years, financial institutions and firms have increasingly adopted synthetic data to address data scarcity and to generate counterfactual market scenarios. However, reproducing all the statistical properties of financial time series, commonly known as stylized facts, remains an open challenge for many existing general-purpose architectures. In this paper, we present a quality-aware generative framework that combines two classes of...

    arxiv.org/abs/2605.27113 · PDF

  21. 21

    Mildly Overparameterized ReLU Networks on Orthogonal Data: Incremental Learning and Implicit Bias

    James Town, Etienne Boursier, Ben Lewis, Matthias Englert, Ranko Lazic

    cs.LG · stat.ML

    The successful training of neural networks hinges on the use of first order optimization methods, yet the theoretical characterization of these methods remains incomplete. This is especially true in settings with mild overparameterization. In this work, we study the gradient flow dynamics of two-layer ReLU networks from small initialization with orthogonal training data. We prove the limiting flow converges to a saddle-to-saddle jump process...

    arxiv.org/abs/2605.27097 · PDF

  22. 22

    Adversarial Dual On-Policy Distillation from Expressive Flow-based Teacher

    Zhenglin Wan, Jingxuan Wu, Xingrui Yu, Chubin Zhang, Mingcong Lei, Bo An, Ivor W. Tsang, Yang You

    cs.LG

    Learning from demonstrations in embodied control is often cast as behavioral cloning, and recent diffusion or flow-matching policies improve this paradigm by modeling multi-modal expert actions. Yet these methods remain offline supervised learners: the policy is trained only on expert states and receives no corrective signal on the states it actually visits. On-policy distillation (OPD) offers a natural remedy, but standard OPD assumes a...

    arxiv.org/abs/2605.27095 · PDF

  23. 23

    ReMoE: Boosting Expert Reuse through Router Fine-Tuning in Memory-Constrained MoE LLM Inference

    Xiongwei Zhu, Xiaojian Liao, Tianyang Jiang, Yusen Zhang, Liang Wang, Limin Xiao

    cs.LG · cs.AI · cs.DC

    Fine-grained Mixture-of-Experts (MoE) models sparsely activate only a subset of experts per token, reducing activated computation while maintaining high model capacity. However, in memory-constrained inference scenarios, only a small set of experts can be cached. Experts not in the cache must be fetched from slow external storage (e.g., UFS), leading to frequent evictions and substantial I/O overhead. We propose ReMoE, a router fine-tuning...

    arxiv.org/abs/2605.27081 · PDF

  24. 24

    Trust Region Q Adjoint Matching

    Yonghoon Dong, Kyungmin Lee, Changyeon Kim, Jaehyuk Kim, Jinwoo Shin

    cs.LG · cs.AI · cs.RO

    Off-policy reinforcement learning of pretrained flow policies remains challenging due to the instability of optimization arising from the multi-step sampling process. Recently, Q-learning with Adjoint Matching (QAM) addressed this issue by reformulating into a memoryless stochastic optimal control (SOC) problem with a learned critic. However, QAM inherits a fundamental fragility of critic-guided improvement: small critic errors are amplified...

    arxiv.org/abs/2605.27079 · PDF

  25. 25

    Two Speeds of Learning: A Representation-Readout Decomposition of Grokking and Double Descent

    Chi-Ning Chou, Oscar Uzdelewicz, Neng-Chun Chiu, Yao-Yuan Yang, SueYeon Chung

    cs.LG · cs.AI

    Training loss and accuracy are the standard signals used to monitor generalization during deep neural network training. Two well-documented phenomena complicate this picture: in grokking, train loss falls rapidly while test performance improves abruptly only after a long delay; in epoch-wise double descent, train loss decreases monotonically while test loss or error rises and falls. Existing accounts are often task-specific, and a...

    arxiv.org/abs/2605.27078 · PDF

  26. 26

    Learning to Orchestrate Agents under Uncertainty

    Mary Chriselda Antony Oliver, Lan Jiang, Aaron Bundi Anampiu, Elaf Almahmoud, Francesco Quinzan, Umang Bhatt

    cs.LG

    Adaptive orchestration of heterogeneous agents requires making sequential delegation decisions under uncertain and evolving agent behaviour, e.g., coordinating specialised AI models with varying reliability, cost, and response quality. While prior work on agent orchestration focuses on performance or cost, uncertainty in agent reliability and output distributions is typically not modelled explicitly at the orchestration level. In this work,...

    arxiv.org/abs/2605.27073 · PDF

  27. 27

    Learning Dynamic Graph Representations through Timespan View Contrasts

    Yiming Xu, Zhen Peng, Bin Shi, Xu Hua, Bo Dong

    cs.LG

    The rich information underlying graphs has inspired further investigation of unsupervised graph representation. Existing studies mainly depend on node features and topological properties within static graphs to create self-supervised signals, neglecting the temporal components carried by real-world graph data, such as timestamps of edges. To overcome this limitation, this paper explores how to model temporal evolution on dynamic graphs...

    arxiv.org/abs/2605.27063 · PDF

  28. 28

    Less is More: Early Stopping Rollout for On-Policy Distillation

    Zhou Ziheng, Jiaqi Li, Huacong Tang, Ying Nian Wu, Demetri Terzopoulos

    cs.LG · cs.AI

    On-policy distillation has recently emerged as a promising alternative to standard sequence-level imitation, training a student by scoring its own rollouts with a teacher model. However, we observe ``Off-policy Teacher Decay'' problem in this paradigm: for the later tokens, with student's earlier trajectory as context that is off-policy to the teacher, the teacher's ability to produce a corrective score would decay, and may fall back to...

    arxiv.org/abs/2605.27028 · PDF

  29. 29

    SQARL: A Size-Agnostic Reinforcement Learning approach for Circuit Allocation in Distributed Quantum Architectures

    Víctor Carballo, Júlia López-Closa, Mario Martin

    cs.LG

    The scaling of quantum processors is currently limited by technical challenges such as decoherence and cross-talk. As the number of qubits grows, interference increases the computational noise. Distributed quantum computing addresses these limitations by interconnecting smaller, easier-to-handle quantum processors (cores), but it introduces the challenge of minimizing slow, error-prone inter-core communication. The task of distributing...

    arxiv.org/abs/2605.27027 · PDF

  30. 30

    SCENT: Aligning Mass Spectra with Molecular Structure for Olfactory Perception

    Ziqi Zhang, Eunyeong Jin, Miguel Vasco, Farzaneh Taleb, Nona Rajabi, Alexandra Gutmann, Jonathan Williams, Antônio...

    cs.LG

    Predicting human olfactory perception from molecular structure has seen remarkable progress, yet these approaches require explicit chemical structure at inference, which is not available in practical sensing settings. We address this gap by exploring direct electron ionization mass spectrometry (EI-MS), a sensing technique that acquires chemically informative fragmentation fingerprints in seconds, as an alternative input modality for...

    arxiv.org/abs/2605.27009 · PDF

  31. 31

    Sampling Data with Chains of Forward-Backward Diffusion Steps

    Hyunmo Kang, Noam Itzhak Levi, Corinna Elena Wegner, Daniel J. Korchinski, Matthieu Wyart

    cs.LG · cond-mat.dis-nn · stat.ML

    Sampling from learned high-dimensional distributions is a foundational computational problem. We introduce U-turn chains: Markov chains obtained by iterating short forward-backward steps of a diffusion model, in which each step proposes a move that remains on the learned data manifold and, paired with a Metropolis-Hastings correction, samples from energy-modified targets. For synthetic languages, we show that minimal U-turn dynamics undergoes...

    arxiv.org/abs/2605.27006 · PDF

  32. 32

    Probabilistic Recurrent Intention Switching Model

    Wenyuan Sheng, Hao Zhu, Joschka Boedecker

    cs.LG · q-bio.NC

    Inverse reinforcement learning (IRL) recovers reward functions from observed behavior, yet traditional methods assume a single stationary reward that cannot capture goal switching within an episode. Recent multi-intention IRL methods address this by segmenting trajectories, but model intention transitions as either a memoryless Markov chain or via manual state augmentation with a fixed history window. We propose the Probabilistic Recurrent...

    arxiv.org/abs/2605.26998 · PDF

  33. 33

    TED: Related Party Transaction guided Tax Evasion Detection on Heterogeneous Graph

    Yiming Xu, Bin Shi, Bo Dong, Jiaxiang Wang, Hua Wei, Qinghua Zheng

    cs.LG

    Tax evasion causes severe losses of government revenues and disturbs the economic order of fair competition. To help alleviate this problem, the latest tax evasion detection solutions utilize expert knowledge to extract features and then train classifiers to determine whether a company is suspected of tax evasion. However, existing solutions mainly focus on the statistical features of the company, but fail to exploit the rich interactive...

    arxiv.org/abs/2605.26984 · PDF

  34. 34

    Convergence of Spectral Descent for Non-smooth Optimization

    Yixuan Yang, Yuqing He, Song Li

    cs.LG · math.OC

    The Muon optimizer has recently demonstrated remarkable empirical success in training large language models. However, the theoretical understanding of its mechanisms remains limited. Current convergence guarantees for Muon rely heavily on smoothness assumptions, leaving its non-smooth convergence behavior largely unexplored. In this work, we take a step toward bridging this gap by investigating Spectral Descent (SD), a simplified variant of...

    arxiv.org/abs/2605.26977 · PDF

  35. 35

    RLVR Datasets and Where to Find Them: Tracing Data Lineage for Better Training Data

    Hsiu-Yuan Huang, Weijie Liu, Chenming Tang, Sanwoo Lee, Kai Yang, Yangkun Chen, Saiyong Yang, Yunfang Wu

    cs.LG

    The proliferation of Reinforcement Learning from Verifiable Rewards (RLVR) datasets has exacerbated provenance collapse due to unclear lineage among existing datasets. To bridge this fragmented RLVR data landscape, we propose Atomic-source Tracing via Lineage-Aware Search (ATLAS), a systematic framework for tracing RLVR datasets back to their atomic sources, attributing over 99.7% of 1.45M instances to 20 atomic sources. Our analysis reveals...

    arxiv.org/abs/2605.26971 · PDF

  36. 36

    When Muon Optimizer Meets Adversarial Training: A Theoretical and Empirical Study

    Jun Yan, Weiquan Huang, Jiankai Zuo, Yujian Mo, Xi Fang, Chengliang Wu, Zeming Wei

    cs.LG

    Adversarial training (AT) remains one of the most reliable empirical defenses against adversarial attacks. Its robustness critically depends on how the underlying min-max objective is optimized. In practice, Stochastic Gradient Descent (SGD) optimizer remains the default optimization choice for AT, whereas adaptive optimizers often improve standard training but may yield inferior robustness. Recently, the Muon optimizer, which orthogonalizes...

    arxiv.org/abs/2605.26929 · PDF

  37. 37

    Agile Online Model Selection: Resolving Adaptation Lag via Safeguarded Large Learning Rates

    Kei Takemura, Ryuta Matsuno, Keita Sakuma

    cs.LG · stat.ML

    Maintaining predictive accuracy in non-stationary environments requires online model selection to adapt autonomously to unknown distribution shifts. However, existing tuning-free algorithms face a fundamental trade-off between robustness and agility. Specifically, to ensure dynamic regret bounds, they must restrict learning rates to small constants (e.g., $O(1)$). This restriction inevitably causes significant adaptation lag during abrupt...

    arxiv.org/abs/2605.26919 · PDF

  38. 38

    EEG-FM-Audit: A Systematic Evaluation and Analysis Pipeline for EEG Foundation Models

    Xianheng Wang, Yige Yang, Damien Coyle

    cs.LG · cs.AI

    Large EEG Foundation Models (FMs) have shown great potential for decoding EEG signals across diverse cognitive tasks. However, existing EEG-FM studies exhibit three critical limitations: opaque supervised baseline tuning, unverified contributions of complex learning paradigms, and a lack of transparency in model decision-making. To address these, we propose EEG-FM-Audit, a comprehensive evaluation and analysis pipeline designed to systematize...

    arxiv.org/abs/2605.26910 · PDF

  39. 39

    SPHERE-JEPA: Spherical Prediction with Homogeneous Embeddings

    Léo Nicollier, Max Dunitz, Marc Pic, Pablo Musé, Enric Meinhardt-Llopis, Gabriele Facciolo

    cs.LG

    A fundamental open question in self-supervised learning (SSL) is the explicit characterization of the optimal geometry of the learned representations. Recently, LeJEPA identified isotropic Gaussian embeddings as optimal for minimizing downstream prediction risk in Euclidean spaces. However, the corresponding problem for distributions supported on lower-dimensional manifolds, such as the hypersphere, remains unexplored. In this work, we...

    arxiv.org/abs/2605.26900 · PDF

  40. 40

    Negligible in Size, Significant in Effect: On Scale Vectors in Large Language Models

    Mingze Wang, Shuchen Zhu, Yuxin Fang, Binghui Li, Kai Shen, Shu Zhong

    cs.LG · cs.AI · stat.ML

    Normalization layers in modern large language models (LLMs) consist of a deterministic normalization operation and a learnable scale vector. While the normalization operation has been extensively studied, the scale vector remains poorly understood despite its ubiquitous use. In this work, we present a systematic study of scale vectors in LLMs from the perspectives of expressivity, optimization, and architectural structure. First, we show...

    arxiv.org/abs/2605.26895 · PDF

  41. 41

    The Strongest Teacher Is Not Always the Best Teacher: Student-Centric Answer Selection

    Zhengyu Hu, Zheyuan Xiao, Linxin Song, Fengqing Jiang, Yutai Li, Zhengyu Chen, Zhihan Xiong, Yue Liu, Junhao Lin,...

    cs.LG · cs.AI · cs.CL

    LLM training increasingly relies on teacher-generated supervision, from synthetic responses to reasoning traces and tool-use demonstrations. Current practice often chooses the highest-performing teacher to generate student training data, implicitly treating teacher test performance as a proxy for teaching quality. We show that this assumption can fail: even when multiple teachers provide correct answers to the same question, the answer from...

    arxiv.org/abs/2605.26872 · PDF

  42. 42

    Generalist Graph Anomaly Detection via Prototype-Based Distillation

    Yiming Xu, Zihan Chen, Zhen Peng, Song Wang, Bin Shi, Bo Dong, Chao Shen

    cs.LG

    Driven by the pressing demand for graph anomaly detection (GAD) in high-stakes domains, the generalist GAD paradigm, which trains a single detector transferable across new graphs, has recently gained growing attention. However, existing methods often rely on scarce and costly annotations for training and sometimes even require few-shot support at inference, which limits their robustness to diverse and unseen anomaly patterns. To address this...

    arxiv.org/abs/2605.26857 · PDF

  43. 43

    RAPNet: Accelerating Algebraic Multigrid with Learned Sparse Corrections

    Yali Fink, Ido Ben-Yair, Lars Ruthotto, Eran Treister

    cs.LG

    The scalable solution of large sparse linear systems is a bottleneck in scientific computing and graph analysis. While algebraic multigrid (AMG) offers optimal linear scaling, its performance is severely constrained by the trade-off between the sparsity and convergence quality of coarse-grid operators. Classical AMG heuristics struggle to balance these objectives, often sacrificing stability or performance for sparsity. We propose RAPNet, a...

    arxiv.org/abs/2605.26854 · PDF

  44. 44

    Learning Energy-Based Models from Stochastic Interpolants using Spatiotemporal Differences

    Hanlin Yu, RuiKang OuYang, Partha Kaushik, Arto Klami, Michael U. Gutmann, Omar Chehab

    cs.LG

    Learning an energy-based model from data samples is a central problem in machine learning. Many recent and popular methods, such as denoising score matching for training energy-based diffusion models, use stochastic interpolants to corrupt data samples at different noise levels indexed by a time variable. This defines a joint density over both the data space and time, and most methods learn its energy through either spatial or temporal...

    arxiv.org/abs/2605.26850 · PDF

  45. 45

    Not All Disagreement Is Learnable: Token Teachability in On-Policy Distillation

    Yuanyi Wang, Su Lu, Yanggan Gu, Pengkai Wang, Yifan Yang, Zhaoyi Yan, Congkai Xie, Jianmin Wu, Hongxia Yang

    cs.LG

    On-policy distillation (OPD) trains a student on its own rollouts with token-level teacher supervision. Recent selective OPD methods exploit the non-uniformity of OPD signals by prioritizing high-entropy or high-disagreement tokens. We revisit this principle and ask: which token-level teacher signals are actually learnable? Using a fixed-context diagnostic that measures same-context teacher-student KL reduction, we show that raw KL...

    arxiv.org/abs/2605.26844 · PDF

  46. 46

    MONA: Muon Optimizer with Nesterov Acceleration for Scalable Language Model Training

    Jiacheng Li, Jianchao Tan, Hongtao Xu, Jiaqi Zhang, Yifan Lu, Yerui Sun, Yuchen Xie, Xunliang Cai

    cs.LG · cs.CL

    The Muon optimizer has recently offered a promising alternative to AdamW for large language model training, leveraging matrix orthogonalization to produce geometry-aware updates. However, like all first-order methods, Muon can become trapped in sharp local minima. In this work, we present MONA, an optimizer that bridges Muon's orthogonalization framework with curvature-aware acceleration. MONA adds an acceleration term directly into Muon's...

    arxiv.org/abs/2605.26842 · PDF

  47. 47

    Periodic Topological Deep Learning for Polymer Design and Discovery

    Yasharth Yadav, Tze Kwang Gerald Er, Atsushi Goto, Kelin Xia

    cs.LG · cs.AI

    Polymers underpin applications across energy, healthcare, and materials science, yet their vast chemical space makes systematic discovery challenging. Most machine learning approaches represent polymers as molecular graphs of a single repeating unit, thereby missing both the periodicity of polymer chains and many-body interactions beyond pairwise bonds. We introduce Periodic-TDL, a deep learning framework built on periodic Vietoris-Rips...

    arxiv.org/abs/2605.26833 · PDF

  48. 48

    The Kalman Evolve: Closing the Gap in Kalman Filtering via Interpretable Algorithm Discovery

    Vasileios Saketos, Ming Xiao

    cs.LG · cs.AI · cs.CV

    State estimation is a fundamental problem in control and signal processing, for which the Kalman Filter provides an optimal solution under linear dynamics, Gaussian noise, and known noise covariances. However, these assumptions often fail in realistic sensing settings such as Doppler radar and LiDAR. In these cases, the optimal estimator is inherently nonlinear, which leads to systematic performance degradation. This creates a performance gap...

    arxiv.org/abs/2605.26830 · PDF

  49. 49

    Innovation: An Almost Characterization of Hallucination

    Nishant P. Das, Piyush Srivastava

    cs.LG · cs.AI · cs.IT

    Hallucination is a central limitation of large language models (LLMs), and substantial effort has been devoted to understanding and mitigating it. Towards this, Kalai and Vempala (STOC 2024) introduced a probabilistic framework formalizing calibration and hallucination, and showed that, with high probability, calibrated LLMs hallucinate roughly at the rate of the "missing mass", a measure of how incomplete the training data is relative to its...

    arxiv.org/abs/2605.26808 · PDF

  50. 50

    PATE-TabTransGAN: Differentially Private Synthetic Tabular Data Generation via Transformer-Based Student Discrimination

    M. Youssef, M. Woźniak

    cs.LG

    Generating high-fidelity synthetic tabular data under formal differential privacy guarantees remains an open challenge. Methods that provide strong theoretical protection typically sacrifice the modeling of inter-feature dependencies required for realistic synthesis, while architectures that excel at capturing complex column relationships offer only empirical privacy guarantees. We present PATE-TabTransGAN, a generative framework that...

    arxiv.org/abs/2605.26802 · PDF

  51. 51

    Latent Recurrent Transformer: Architecture Exploration, Training Strategies, and Scaling Behavior

    Zeyi Huang, Xuehai He, LiLiang Ren, Yiping Wang, Baolin Peng, Hao Cheng, Shuohang Wang, Pengcheng He, Jianfeng Gao,...

    cs.LG · cs.CL

    We study Latent Recurrent Transformer (LRT), a lightweight augmentation of autoregressive transformers that reuses a high-level source-layer hidden state from the previous token as recurrent memory for the next token. Because this source state is already computed during ordinary decoding, LRT adds a cross-layer recurrent latent pathway across positions without inserting pause tokens or extra depth loops, and the standard attention mechanism...

    arxiv.org/abs/2605.26797 · PDF

  52. 52

    Pretrained Approximators for Low-Thrust Trajectory Cost and Reachability

    Zhong Zhang, Giacomo Acciarini, Dario Izzo, Hexi Baoyin, Francesco Topputo

    cs.LG · physics.space-ph

    Low-thrust trajectory design relies heavily on repeated evaluations of fuel consumption and transfer feasibility, which require expensive optimal control solutions. In this work, we show these quantities can be accurately approximated by machine learning surrogates, enabling fast and scalable evaluation across a wide range of scenarios. By increasing both dataset size and model capacity, we observe that low-thrust trajectory optimization...

    arxiv.org/abs/2605.26790 · PDF

  53. 53

    Ratio-Variance Regularized Policy Optimization

    Yu Luo, Shuo Han, Yihan Hu, Lei Lv, Huaping Liu, Fuchun Sun, Jianye Hao, Dong Li

    cs.LG · cs.AI

    Standard on-policy reinforcement learning relies on heuristic clipping to enforce trust regions, but this mechanism imposes a severe cost by indiscriminately truncating high-return yet high-divergence updates. We demonstrate that explicitly constraining the policy ratio variance provides a principled local approximation to trust-region constraints, eliminating the need for binary hard clipping. By acting as a distributional ``soft brake'',...

    arxiv.org/abs/2605.26784 · PDF

  54. 54

    Towards Generalization-Oriented Models for Vehicle Routing Problems with Mixture-of-Experts

    Changhao Miao, Yuntian Zhang, Tongyu Wu, Fang Deng, Chen Chen

    cs.LG · cs.AI

    In recent years, Deep Reinforcement Learning (DRL) has achieved substantial progress on Vehicle Routing Problems (VRPs). However, existing DRL-based methods are typically trained on instances generated from a uniform distribution, which limits their performance under real-world distribution shifts. In this paper, we aim to develop a generalization-oriented model that partitions the policy network into multiple modules and adaptively...

    arxiv.org/abs/2605.26776 · PDF

  55. 55

    Adversarial Training for Robust Coverage Network under Worst-case Facility Losses

    Changhao Miao, Yuntian Zhang, Tongyu Wu, Fang Deng, Chen Chen

    cs.LG · cs.AI

    The Maximal Covering Location-Interdiction Problem (MCLIP) is a classic bi-level optimization problem, which is fundamental to resilient infrastructure planning yet remains computationally intractable. Specifically, the upper level determines facility locations to maximize coverage, while the lower level executes worst-case interdiction to minimize the coverage. The strong coupling between the upper and lower levels, combined with their...

    arxiv.org/abs/2605.26763 · PDF

  56. 56

    Time Series Causal Discovery via Context-Conditioned and Causality-Augmented Pretraining

    Biao Ouyang, Tengxue Zhang, Zhihao Zhuang, Yang Shu, Chenjuan Guo, Bin Yang

    cs.LG

    Causal discovery from time series is critical for many real-world applications, such as tracing the root causes of anomalies. Existing approaches typically rely on dataset-specific optimization, making it difficult to transfer their causal discovery capabilities to new time series governed by diverse causal mechanisms. In this paper, we propose \textbf{PTCD}, a novel \textbf{P}retraining framework for \textbf{T}ime-series \textbf{C}ausal...

    arxiv.org/abs/2605.26759 · PDF

  57. 57

    Localizing Memorized Regions in Diffusion Models via Coordinate-Wise Curvature Differences

    Gwangho Kim, Sungyoon Lee

    cs.LG

    Diffusion models can unintentionally memorize training samples, raising concerns about privacy and copyright. While recent methods can detect memorization, they often rely on global or model-specific signals and provide limited insight into where memorization appears within a generated image. We provide a geometric characterization of local memorization as a coordinate-wise variance collapse. However, such collapse can also arise from...

    arxiv.org/abs/2605.26756 · PDF

  58. 58

    Stabilizing Recurrent Dynamics for Test-Time Scalable Latent Reasoning in Looped Language Models

    Xiao-Wen Yang, Ziyu Han, Xi-Hua Zhang, Wen-Da Wei, Jie-Jing Shao, Lan-Zhe Guo, Yu-Feng Li

    cs.LG · cs.AI

    Looped Language Models (LoopLMs) enable efficient latent reasoning through depth recurrence, yet exhibit unreliable test-time scaling behavior: performance often peaks at a certain iteration depth and then collapses with further recurrence. Through latent dynamics analysis, we find an inherent trade-off between stability and effectiveness in existing architectures and strategies. By conceptualizing reasoning as uncertainty reduction, we...

    arxiv.org/abs/2605.26733 · PDF

  59. 59

    APEX: Amplitude Anchors and Phase Priors for Target-Scarce Higher-Frequency Wave Prediction

    Yifan Sun, Lei Cheng, Sijie Chen, Ting Zhang, Jianlong Li, Shikai Fang

    cs.LG

    Learning-based surrogates have become increasingly effective for wave-field prediction, and neural operators in particular have shown strong performance within observed frequency regimes. However, higher-frequency prediction under scarce target supervision remains comparatively underexplored, especially in wave problems where higher-frequency data are substantially more expensive to simulate or measure than lower-frequency data. A central...

    arxiv.org/abs/2605.26732 · PDF

  60. 60

    MTL-FNO: A Lightweight Multi-Task Fourier Neural Operator for Sparse Field Reconstruction

    Siyu Ye, Shihang Li, Zhiqiang Gong, Benrong Zhang, Weien Zhou, Yiyong Huang, Wen Yao

    cs.LG

    Efficient onboard multi-field sparse reconstruction is essential for the autonomous operation of aerospace vehicles. While existing deep learning models exhibit promise for single-field reconstruction, deploying multiple independent models leads to prohibitive model size growth and fails to exploit cross-field correlations, particularly under few-shot conditions. To address these challenges, we first propose a lightweight multi-task Fourier...

    arxiv.org/abs/2605.26718 · PDF

  61. 61

    Image Feature Fusion-based Federated Client Unlearning (FCU)

    Hangyi Shen, Yizhi Pan, Tiansuo Li, Weiqi Jiang, Guanqun Sun

    cs.LG

    Major data protection regulations all mention the "right to be forgotten," and that's what pushed federated unlearning (FU) techniques forward. But one stubborn issue remains: catastrophic forgetting--you erase the target knowledge, yet somehow you also end up throwing out essential retained knowledge, which then hurts the model's global generalization. To get a better balance between unlearning effectiveness and generalization ability, we...

    arxiv.org/abs/2605.26715 · PDF

  62. 62

    SL-BiLEM: Structured Learnable Behavior-in-the-Loop Epidemic Modeling for Forecasting and Policy Evaluation

    Haochun Wang, Sendong Zhao, Jingbo Wang, Yanrui Du, Bing Qin, Ting Liu

    cs.LG · cs.AI

    Epidemic forecasting faces a fundamental challenge: human behavior dynamically responds to disease spread, creating feedback loops that induce distribution shifts at policy intervention points. This renders data-driven models unreliable under distribution shift. We propose \textbf{SL-BiLEM} (Structured Learnable Behavior-in-the-Loop Epidemic Model), leveraging physical constraints as regularization for robust extrapolation. The framework...

    arxiv.org/abs/2605.26704 · PDF

  63. 63

    Model Merging on Loss Landscape: A Geometry Perspective

    Juanwu Lu, Anand Bhaskar, Brian Axelrod, Ekaterina Tolstaya, Tristan Emrich

    cs.LG · cs.AI · stat.ML

    Model merging offers a promising avenue for knowledge integration and parallel development without retraining. Yet, existing methods either ignore the geometry of the loss landscape or rely on intractable full-space Hessian approximations. We propose EpiMer, a framework that casts model merging as solving the Fréchet mean on a Riemannian manifold and restricts the computation to a low-rank subspace spanned by the task vectors. With the...

    arxiv.org/abs/2605.26693 · PDF

  64. 64

    Self-Improvement Imitation with Biologically Guided Search for Protein Design Under Oracle Budgets

    Ashima Khanna, Dominik Grimm

    cs.LG · cs.AI · q-bio.QM

    Protein sequence optimization under tight oracle budgets requires methods that explore vast combinatorial spaces while making each evaluation informative. Existing reinforcement learning and off-policy generative approaches often degrade under surrogate noise, and position-agnostic mutation proposals risk disrupting functionally critical residues. We introduce SILO, a trajectory-level self-improvement imitation framework for oracle-budgeted...

    arxiv.org/abs/2605.26690 · PDF

  65. 65

    Beyond Trajectory-Level Attribution: Graph-Based Credit Assignment for Agentic Reinforcement Learning

    Xin Cheng, Shuo He, Lang Feng, HaiYang Xu, Ming Yan, Lei Feng, Bo An

    cs.LG · cs.AI

    Group-based reinforcement learning (RL) methods have achieved remarkable success in improving the performance of large language models (LLMs) and have been rapidly extended to agentic tasks. However, their credit assignment relies heavily on coarse-grained trajectory-level attribution according to final outcomes, making it difficult to capture the contribution of individual steps, such as valuable steps obscured within failed trajectories. To...

    arxiv.org/abs/2605.26684 · PDF

  66. 66

    WINDQuant: Weight-Informed Neural Decision-Making for Global Mixed-Precision LLM Quantization

    Phong Nam Huu Nguyen, Khoi M. Le, Cong-Duy T Nguyen, Anh Tuan Luu, Thong Thanh Nguyen, Tho Quan

    cs.LG

    Quantization is an effective approach to reduce the memory footprint and inference cost of large language models (LLMs), yet maintaining performance in the ultra-low-bit regime remains challenging. Existing post-training methods often suffer from severe accuracy degradation, while quantization-aware training requires costly retraining and additional resources. Moreover, most mixed-precision strategies rely on coarse-grained or heuristic...

    arxiv.org/abs/2605.26660 · PDF

This edition is part of The Daily Abstract — cs.LG archive. Subscribe to receive these in your inbox each morning, automatically translated to Spanish, with reply-to-PDF: arxivdaily.ignorelist.com.

Colophon Set in Georgia, with system sans for interface chrome and a monospaced stack for code and paper identifiers. Sole accent: amber #D99C5E. Built and served on an always-free VM. The masthead is set 14% letterspaced because newspapers do that and it works.