Process-Level Supervision

Follow this topic Ask about this topic

Momentum

33 papers in the last four weeks, up 200% on the four weeks before. 0.3% of all new papers.

Jul 13Week of Sep 28

Latest papers 154

All topics

Spatial Blindness in Whole-Slide Multiple Instance Learning
May 17, 2026Xiangyu Li, Ran SuWhole-Slide Images Multiple Instance Learning
College of Intelligence and Computing Tianjin University Tianjin, China
SADP: Subgoal-Aware Diffusion Policy for Explainable Robots Learned from Foundation Model Generated Demonstrations
May 16, 2026Site Hu, Takato HoriiDiffusion Policies Robot Systems
Department of Systems Innovation, Graduate School of Engineering Science, Osaka University, Osaka, Japan
SceneForge: Structured World Supervision from 3D Interventions
May 14, 2026Jizhizi Li, Jiayang Ao, Danny Wicks +1Scene Reasoning Multimodal Learning
1Canva Research
Hessian Matching for Machine-Learned Coarse-Grained Molecular Dynamics
May 12, 2026Sanya Murdeshwar, Sanjit Shashi, Kevin Bachelor +3Molecular Dynamics Protein Structure Prediction
University of California, Santa Cruz · GiwoTech Inc. · Pennsylvania State University
StepCodeReasoner: Aligning Code Reasoning with Stepwise Execution Traces via Reinforcement Learning
May 12, 2026Hao Wang, Rui Li, Lei Sha +1Code Generation Search-Augmented Reasoning
Beihang University, Beijing, China · Peking University, Beijing, China · Zhongguancun Laboratory, Beijing, China +1
Rethinking Supervision Granularity: Segment-Level Learning for LLM-Based Theorem Proving
May 12, 2026Shuo Xu, Jiakun Zhang, Junyu Lai +2Theorem Proving Process-Level Supervision
State Key Laboratory for Novel Software Technology, Nanjing University, China
A Controlled Counterexample to Strong Proxy-Based Explanations of OOD Performance: in a Fixed Pretraining-and-Probing Setup
May 12, 2026Hongmin LiGeneration Provenance Out-Of-Distribution
School of Life Science and Technology, Institute of Science Tokyo · Department of Computational Biology and Medical Sciences, Graduate School of Frontier Sciences, The University of Tokyo
PathISE: Learning Informative Path Supervision for Knowledge Graph Question Answering
May 11, 2026Shengxiang Gao, Chao Lei, Jey Han Lau +1Knowledge Graphs Question
The University of Melbourne
Modeling Atomic Conformational Ensembles of Proteins via Test-Time Supervision of Boltz-2 on Cryo-EM Density Maps
May 11, 2026Jay Shenoy, Miro Astore, Axel Levy +3Cryo-Electron Microscopy Conformational Ensembles
Department of Computer Science Stanford University, Flatiron Institute, & SLAC National Lab Stanford, CA 94305, USA · Center for Computational Biology & Center for Computational Mathematics Flatiron Institute New York, NY 10010, USA · Department of Electrical Engineering Stanford University & SLAC National Lab Stanford, CA 94305, USA +2
Think-with-Rubrics: From External Evaluator to Internal Reasoning Guidance
May 8, 2026Jiachen Yu, Zhihao Xu, Junjie Wang +1Process-Level Supervision Reasoning Traces
1Tsinghua University · 2Renmin University of China
Contrastive Identification and Generation in the Limit
May 7, 2026Xiaoyu Li, Andi Han, Jiaojiao Jiang +1Contrastive Learning Adversarial Corruption
University of New South Wales · University of Sydney
VISD: Enhancing Video Reasoning via Structured Self-Distillation
May 7, 2026Hao Lin, Kunyang Lv, Xu Jiang +5Latent Visual Reasoning Process-Level Supervision
1HUST · 2Wuhan University · 3Peking University +1
Knowledge-Graph Paths as Intermediate Supervision for Self-Evolving Search Agents
May 7, 2026Huyu Wu, Jun Liu, Xiaochi Wei +3Self-Evolving Agents Search Agents
Xiaohongshu Inc., Beijing, China
Self-Induced Outcome Potential: Turn-Level Credit Assignment for Agents without Verifiers
May 6, 2026Senkang Hu, Yong Dai, Xudong Han +4Credit Assignment Process-Level Supervision
Hong Kong JC STEM Lab of Smart City · City University of Hong Kong · University of Oxford +4
Every Step Counts: Step-Level Credit Assignment for Tool-Integrated Text-to-SQL
May 6, 2026Yaxun Dai, Baolin Sun, Junying Wang +6Text-To-Sql Credit Assignment
Institute of Computer Science and Technology, Soochow University, China · Ant Digital Technologies, Ant Group · School of Management, University of Science and Technology of China
A Workflow-Oriented Framework for Asynchronous Human-AI Collaboration in Hybrid and Compute-Intensive HPC Environments
May 5, 2026Sergio Mendoza, Cedric Bhihe, Natalia Zamora +5High-Performance Computing Human-Ai Collaboration
1Barcelona Supercomputing Center, Spain · 2NTT DATA, Spain
Verifiable Counterfactual Supervision for Process Reward Models
May 4, 2026Yinghui Chi, Yuanhong WangProcess Reward Model Counterfactual Learning
Jilin University
JURY-RL: Votes Propose, Proofs Dispose for Label-Free RLVR
Apr 28, 2026Xinjie Chen, Biao Fu, Jing Wu +4Reinforcement Learning With Verifiable Reward Verifiable Rewards
Tongyi Lab, Alibaba Group, Hangzhou, China
Learning to Think from Multiple Thinkers
Apr 27, 2026Nirmit Joshi, Roey Magen, Nathan Srebro +2Learnability Chain-of-Thought Reasoning
Toyota Technological Institute at Chicago. · Weizmann Institute of Science. · New York University.
Negative Ontology of True Target for Machine Learning: Towards Evaluation and Learning under Democratic Supervision
Apr 27, 2026Yongquan YangProcess-Level Supervision Democracy
Institute of Sciences for AI, Chengdu, Sichuan, China
GRPO-VPS: Enhancing Group Relative Policy Optimization with Verifiable Process Supervision for Effective Reasoning
Apr 22, 2026Jingyi Wang, Lei Zhu, Tengjin Weng +8Reinforcement Learning With Verifiable Reward Better Call Group Relative Policy Optimization
Tsinghua University · Huawei Noah’s Ark Lab · Shenzhen University
Can Explicit Physical Feasibility Benefit VLA Learning? An Empirical Study
Apr 20, 2026Yubai Wei, Chen Wu, Hashem HaghbayanVision-Language-Action Framework Photometric Supervision
Department of Computing, University of Turku, Finland
Internalizing Outcome Supervision into Process Supervision: A New Paradigm for Reinforcement Learning for Reasoning
Apr 19, 2026Fei Ding, Yongkang Zhang, Runhao Liu +4Process Reward Model Process-Level Supervision
Alibaba Group · Tsinghua University
ACE: Self-Evolving LLM Coding Framework via Adversarial Unit Test Generation and Preference Optimization
Apr 17, 2026Yixu Huang, Xinglei Yu, Zhongyu WeiCode Generation Adversarial Evaluation
School of Data Science Fudan University Shanghai, China
Hybrid Decision Making via Conformal VLM-generated Guidance
Apr 16, 2026Debodeep Banerjee, Burcu Sayin, Stefano Teso +1Large Language Model Decisions Guidance
DI, University of Pisa · DISI, University of Trento · CIMeC, university of Trento
BioELX: Context-Aware Cross-lingual Biomedical Entity Linking without Task-Specific Supervision
Apr 9, 2026Yi Wang, Corina Dima, Liangyu Zhong +1Medical Ontology Life Sciences Research
Institute for AI, University of Stuttgart · Technical University of Berlin · University of Southampton
RAWR: Reward Assignment Without Rollouts in Verifiable Domains
Mar 18, 2026Corentin Royer, Anna Hedström, Debarun Bhattacharjya +3Process Reward Model Reinforcement Learning With Verifiable Reward
International Business Machines · Department of Computer Science, ETH Zurich, 8092 Zurich, Switzerland · ETH AI Center +2
SCRIBE: Structured Mid-Level Supervision for Tool-Using Language Models
Jan 7, 2026Yuxuan Jiang, Francis FerraroProcess Reward Model Process-Level Supervision
University of Maryland, Baltimore County
Deductive Logic in Language Models: Horizontal vs Vertical Reasoning
Oct 10, 2025Davide Maltoni, Matteo FerraraDeduction Small Large Language Models
Department of Computer Science and Engineering, University of Bologna, Italy
CPATTA: Conformal Supervision Allocation For Active Test-Time Adaptation
Sep 30, 2025Tingyu Shi, Fan Lyu, Haihua Zhu +2Stable Test-Time Adaptation Online Conformal Prediction
Computer Science and Engineering, University of California San Diego · New Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences · College of Computer Science and Electronic Engineering, Hunan University
Neuro-symbolic Weak Supervision: Theory and Semantics
Mar 24, 2025Nijesh Upreti, Vaishak BelleNeuro-Symbolic Framework Multiple Instance Learning
School of Informatics, University of Edinburgh, Edinburgh, UK
From My View to Yours: Learning Egocentric Cues from Exocentric Video using Privileged Egocentric Supervision
Jan 10, 2025Dominick Reilly, Manish Kumar Govind, Le Xue +1Egocentric Video Egocentric Vision
University of North Carolina at Charlotte · Elorian AI
MERIT: Matching Expertise via Rubric-Informed Training for Reviewer Assignment
Date pendingZixuan Yang, Yibo Zhao, Weicong Liu +1Reviewer Task-Specific Rubrics
School of Data Science and Engineering, East China Normal University
A Recipe for Long-Context Reasoning in Large Language Models via On-Policy Optimization and Distillation
Date pendingMiguel Moura Ramos, Duarte M. Alves, André F. T. MartinsEfficient Long-Context Inference Long-Context Reasoning
Instituto Superior Técnico, Universidade de Lisboa · Instituto de Telecomunicações · TransPerfect +1

May 17, 2026, cs.CVJ/K move · Enter open · S save

Spatial Blindness in Whole-Slide Multiple Instance Learning

Xiangyu Li, Ran Su

College of Intelligence and Computing Tianjin University Tianjin, China

Whole-Slide Images Multiple Instance Learning Spatial Blindness Tissue Visual Embeddings Spatial Intent+1

Open paper AskPDF arXiv

Whole-slide MIL models are often called context-aware once graphs, Transform ers, or state-space modules are placed above patch embeddings. We show that this label can be deceptive. On pathology tasks where tissue architecture is part of the diagnostic signal, several strong MIL baselines retain nearly unchanged slide level AUC after patch coordinates are permuted. Their predictions are accurate, but largely compositional. We refer to this failure mode as spatial blindness. Our explanation is optimization-based: dense appearance statistics are learned early under slide-level supervision, leaving weak gradients for sparse spatial relations. ResTopoMIL addresses the issue by first fitting a permutation-invariant prototype histogram and then freezing it while a lightweight graph branch learns the residual under a coordinate-shuffling constraint. The architecture is simple by design; the intervention is in how the spatial branch is trained. Across 9 public WSI bench marks, ResTopoMIL improves classification and survival prediction with 1.15M parameters, restores sensitivity to coordinate perturbation, and gives stronger lo calization evidence on CAMELYON-16.

Momentum

Related topics

Latest papers 154