Isolater - Feed

Ax Yirong Zeng, Xiao Ding, Yufei Liu, Yuxian Wang, Qunyao Du, Yutai Hou, Wu Ning, Haonan Song, Duyu Tang, Dandan Tu, Bing Qin, Ting Liu 3/17/2026

AutoTool: Automatic Scaling of Tool-Use Capabilities in RL via Decoupled Entropy Constraints

RL method for automatically scaling tool-use capabilities in agents through decoupled entropy constraints addressing overthinking.

Ax Heejin Jo 3/17/2026

Prompt Complexity Dilutes Structured Reasoning: A Follow-Up Study on the Car Wash Problem

Study examining whether STAR reasoning prompt technique maintains effectiveness when embedded in complex production system prompts.

Ax Bakhtawar Ahtisham, Kirk Vanacore, Rene F. Kizilcec 3/17/2026

Optimizing LLM Annotation of Classroom Discourse through Multi-Agent Orchestration

Multi-agent LLM orchestration for annotating educational classroom discourse with improved consistency over single-pass annotation.

Ax Majid Ghasemi, Mark Crowley 3/17/2026

Learning When to Trust in Contextual Bandits

Analysis of contextual bias in reinforcement learning feedback sources; proposes methods addressing selective truthfulness in evaluators.

Ax Rishab Alagharu, Ishneet Sukhvinder Singh, Shaibi Shamsudeen, Zhen Wu, Ashwinee Panda 3/17/2026

From Refusal Tokens to Refusal Control: Discovering and Steering Category-Specific Refusal Directions

Technique to discover and steer category-specific refusal directions in language models for fine-grained safety control at inference time.

Ax Sahar Vahdati, Andrei Aioanei, Haridhra Suresh, Jens Lehmann 3/17/2026

The ARC of Progress towards AGI: A Living Survey of Abstraction and Reasoning

Survey analyzing 82 approaches across ARC-AGI benchmark versions; documents consistent 2-3x performance degradation across paradigms.

Ax Jaroslaw Hryszko 3/17/2026

Do Large Language Models Get Caught in Hofstadter-Mobius Loops?

Examines whether RLHF-trained LLMs face contradictory training objectives analogous to Hofstadter-Mobius loops.

Ax Gideon Popoola, John Sheppard 3/17/2026

MESD: Detecting and Mitigating Procedural Bias in Intersectional Groups

Metric for detecting procedural bias in machine learning models across intersectional demographic groups.

Ax Shiva Krishna Reddy Malay, Shravan Nayak, Jishnu Sethumadhavan Nair, Sagar Davasam, Aman Tiwari, Sathwik Tejaswi Madhusudhan, Sridhar Krishna Nemala, Srinivas Sunkara, Sai Rajeswar 3/17/2026

EnterpriseOps-Gym: Environments and Evaluations for Stateful Agentic Planning and Tool Use in Enterprise Settings

Benchmark environment for evaluating AI agents in enterprise workflows with long-horizon planning, persistent state management, and access controls.

Ax Rana Shahout, Hayder Tirmazi, Minlan Yu, Michael Mitzenmacher 3/17/2026

Orla: A Library for Serving LLM-Based Multi-Agent Systems

Open source library for constructing and serving LLM-based multi-agent systems with workflow orchestration and tool execution abstraction.

Ax Son Nguyen, Xinyuan Liu, Ransalu Senanayake 3/17/2026

LLM Routing as Reasoning: A MaxSAT View

Constraint-based approach formulating LLM routing as MaxSAT problem to match queries with appropriate models based on natural language preferences.

Ax Sasank Annapureddy, John Mulcahy, Anjaneya Prasad Thamatani 3/17/2026

StatePlane: A Cognitive State Plane for Long-Horizon AI Systems Under Bounded Context

Framework addressing context window limitations in LLMs through dynamic cognitive state management for long-horizon agent tasks.

Ax Mingchen Shao, Yuzhang Xie, Carl Yang, Jiaying Lu 3/17/2026

LLM-MINE: Large Language Model based Alzheimer's Disease and Related Dementias Phenotypes Mining from Clinical Notes

LLM-based system for extracting Alzheimer's disease phenotypes from clinical notes in electronic health records.

Ax Zhihao Chen, Jiahui Wang, Yizhou Chen, Xiaozhong Ji, Xiaobin Hu, Jimin Hong, Wolfram Andreas Bosbach, Axel Rominger, Ali Afshar-Oromieh, Hongming Shan, Kuangyu Shi 3/17/2026

TheraAgent: Multi-Agent Framework with Self-Evolving Memory and Evidence-Calibrated Reasoning for PET Theranostics

Multi-agent framework using LLMs with self-evolving memory for predicting treatment outcomes in PET theranostics for prostate cancer.

Ax Xiao Wang, Lu Dong, Ifeoma Nwogu, Srirangaraj Setlur, Venu Govindaraju 3/17/2026

InterventionLens: A Multi-Agent Framework for Detecting ASD Intervention Strategies in Parent-Child Shared Reading

Multi-agent system using LLMs to automatically detect autism intervention strategies in parent-child reading interactions from video.

Ax Qizhao Jin, Xianhuang Xu, Yong Cao, Shiming Xiang, Xinyu Xiao 3/17/2026

MeTok: An Efficient Meteorological Tokenization with Hyper-Aligned Group Learning for Precipitation Nowcasting

Transformer-based tokenization method for precipitation nowcasting emphasizing synergistic interactions between meteorological elements.

Ax Dingding Cao, Mieow Kee Chan, Wan Sieng Yeo, Said Bey, Alberto Figoli 3/17/2026

Intelligent Materials Modelling: Large Language Models Versus Partial Least Squares Regression for Predicting Polysulfone Membrane Mechanical Performance

Benchmarking LLMs against traditional regression for predicting polysulfone membrane mechanical properties from limited experimental data.

Ax Yiling Tao, Xinran Zheng, Shuo Yang, Meiling Tao, Xingjun Wang 3/17/2026

GroupGuard: A Framework for Modeling and Defending Collusive Attacks in Multi-Agent Systems

GroupGuard framework modeling and defending against collusive attacks where multiple LLM agents coordinate to mislead systems.

Ax Tuoshi Qi, Shenshen Bu, Yingfei Xiang, Zhiming Dai 3/17/2026

EviAgent: Evidence-Driven Agent for Radiology Report Generation

Evidence-driven agent framework for radiology report generation using multimodal LLMs with explicit visual evidence tracing and clinical reasoning.

Ax Suhwan Choi, Yunsung Lee, Yubeen Park, Chris Dongjoo Kim, Ranjay Krishna, Dieter Fox, Youngjae Yu 3/17/2026

vla-eval: A Unified Evaluation Harness for Vision-Language-Action Models

Open source evaluation harness for vision-language-action models decoupling inference from benchmarks via WebSocket protocol and Docker isolation.

Ax Haitao Jiang, Wenbo Zhang, Jiarui Yao, Hengrui Cai, Sheng Wang, Rui Song 3/17/2026

Supervised Fine-Tuning versus Reinforcement Learning: A Study of Post-Training Methods for Large Language Models

Comparative study of supervised fine-tuning versus reinforcement learning as post-training methods for LLMs, examining theoretical connections.

Ax Halimat Afolabi, Zainab Afolabi, Elizabeth Friel, Jude Roberts, Antonio Ji-Xu, Lloyd Chen, Egheosa Ogbomo, Emiliomo Imevbore, Phil Eneje, Wissal El Ouahidi, Aaron Sohal, Alisa Kennan, Shreya Srivastava, Anirudh Vairavan, Laura Napitu, Katie McClure 3/17/2026

Faithful or Just Plausible? Evaluating the Faithfulness of Closed-Source LLMs in Medical Reasoning

Black-box evaluation of faithfulness in medical reasoning explanations across closed-source LLMs like ChatGPT and Gemini.

Ax Mario Heidrich, Jeffrey Heidemann, R\"udiger Buchkremer, Gonzalo Wandosell Fern\'andez de Bobadilla 3/17/2026

A Systematic Evaluation Protocol of Graph-Derived Signals for Tabular Machine Learning

Systematic evaluation protocol assessing reliability and robustness of graph-derived signals in tabular machine learning.

Ax Zhijie Wang 3/17/2026

GRPO and Reflection Reward for Mathematical Reasoning in Large Language Models

GRPO and reflection reward methods for improving mathematical reasoning in LLMs through proactive reflection encouragement during training.

Ax Raj Navakoti, Saideep Navakoti 3/17/2026

Demand-Driven Context: A Methodology for Building Enterprise Knowledge Bases Through Agent Failure

Demand-driven methodology for building enterprise knowledge bases by analyzing LLM agent failures to capture domain-specific tribal knowledge.

Ax Vishnu Narayanan Anilkumar, Abhijith Sreesylesh Babu, Trieu Hai Vo, Mohankrishna Kolla, Alexander Cuneo 3/17/2026

Relationship-Aware Safety Unlearning for Multimodal LLMs

Framework for relationship-aware safety unlearning in multimodal LLMs, addressing relational safety failures without collateral damage.

Ax Yanqi Pan, Qinghao Huang, Weihao Yang 3/17/2026

Memory as Asset: From Agent-centric to Human-centric Memory Management

Memory-as-Asset paradigm for human-centric personal memory management as prerequisite for extending LLM knowledge boundaries through self-evolution.

Ax Kirushikesh D B, Manish Kesarwani, Nishtha Madaan, Sameep Mehta, Aldrin Dennis, Siddarth Ajay, Rakesh B R, Renu Rajagopal, Sudheesh Kairali 3/17/2026

Agentic DAG-Orchestrated Planner Framework for Multi-Modal, Multi-Hop Question Answering in Hybrid Data Lakes

DAG-orchestrated agentic planner for multi-hop question answering over hybrid data lakes combining structured and unstructured sources.

Ax Mohamed Aghzal, Gregory J. Stein, Ziyu Yao 3/17/2026

Why Do LLM-based Web Agents Fail? A Hierarchical Planning Perspective

Hierarchical planning framework for analyzing LLM web agent failures across planning, execution, and replanning layers for process-based evaluation.

Ax Fiona Y. Wang, Lee Marom, Subhadeep Pal, Rachel K. Luu, Wei Lu, Jaime A. Berkovich, Markus J. Buehler 3/17/2026

Autonomous Agents Coordinating Distributed Discovery Through Emergent Artifact Exchange

ScienceClaw + Infinite framework enabling autonomous agents to conduct scientific research through emergent artifact exchange and 300+ interoperable skills.

Ax Sagi Ohayon, Boaz Taitler, Omer Ben-Porat 3/17/2026

Contests with Spillovers: Incentivizing Content Creation with GenAI

Economics paper analyzing spillovers and incentive structures in content creation enabled by GenAI reuse capabilities.

Ax Tiantian Mi, Dongming Shan, Zhen Huang, Yiwei Qin, Muhang Xie, Yuxuan Qiao, Yixiu Liu, Chenyang Zhou, Pengfei Liu 3/17/2026

Data Darwinism Part II: DataEvolve -- AI can Autonomously Evolve Pretraining Data Curation

Framework for autonomous evolution of data curation strategies across heterogeneous pretraining categories, extending Data Darwinism hierarchy to multi-domain scenarios.

Ax Shengda Fan, Xuyan Ye, Yupeng Huo, Zhi-Yuan Chen, Yiju Guo, Shenzhi Yang, Wenkai Yang, Shuqi Ye, Jingwen Chen, Haotian Chen, Xin Cong, Yankai Lin 3/17/2026

AgentProcessBench: Diagnosing Step-Level Process Quality in Tool-Using Agents

Benchmark for evaluating step-level process quality in tool-using LLM agents, addressing brittleness in long-horizon interactions with irreversible side effects.

Ax Ying Xie 3/17/2026

Learning to Forget: Sleep-Inspired Memory Consolidation for Resolving Proactive Interference in Large Language Models

Method for resolving proactive interference in LLMs using sleep-inspired memory consolidation to improve context window retrieval accuracy.

Ax Diego Ezequiel Cervera 3/17/2026

Expert Mind: A Retrieval-Augmented Architecture for Expert Knowledge Preservation in the Energy Sector

System using RAG and LLMs to preserve and query tacit expert knowledge in industrial organizations through structured multimodal capture.

Ax Mayank Vyaas, Abhijit Chakrabroty, Vivek Gupta 3/17/2026

JobMatchAI An Intelligent Job Matching Platform Using Knowledge Graphs, Semantic Search and Explainable AI

Production-ready job matching platform using transformer embeddings, skill knowledge graphs, and interpretable reranking for candidate matching.

Ax Varun Pratap Bhardwaj 3/17/2026

SuperLocalMemory V3: Information-Geometric Foundations for Zero-LLM Enterprise Agent Memory

Information-geometric foundations for persistent memory in AI agents, covering retrieval metrics, lifecycle management, and consistency detection.

Ax Yaofang Zhang, Adnan Darwiche 3/17/2026

Scaling the Explanation of Multi-Class Bayesian Network Classifiers

Algorithm for compiling Bayesian network classifiers into logical formulas for explainability, extending prior work beyond binary classification.

Ax Adam Dejl, Matthew Williams, Francesca Toni 3/17/2026

Argumentation for Explainable and Globally Contestable Decision Support with LLMs

Research on augmenting LLMs with computational argumentation to provide explainable decisions and enable contestation of outputs in high-stakes domains.

Ax Thuy Ngoc Nguyen, Duy Nhat Phan, Cleotilde Gonzalez 3/17/2026

Dynamic Theory of Mind as a Temporal Memory Problem: Evidence from Large Language Models

Investigation of dynamic Theory of Mind in LLMs as temporal memory problem, measuring ability to represent and update beliefs over time.

Ax Mark Baciak, Thomas A. Cellucci, Deanna M. Falkowski 3/17/2026

Punctuated Equilibria in Artificial Intelligence: The Institutional Scaling Law and the Speciation of Sovereign AI

Theoretical framework applying punctuated equilibrium to AI development, challenging continuous progress assumptions and predicting capability phase transitions.

Ax J Rosser 3/17/2026

Gradient Atoms: Unsupervised Discovery, Attribution and Steering of Model Behaviors via Sparse Decomposition of Training Gradients

Gradient Atoms method for unsupervised discovery of model behaviors via sparse decomposition of training gradients without requiring query behaviors.

Ax JooHyun Park, HyeongYeop Kang 3/17/2026