Isolater - Feed

Ax Maohao Shen, Kaiwen Zha, Zexue He, Zhang-Wei Hong, Siru Ouyang, J. Jon Ryu, Prasanna Sattigeri, Suhas Diggavi, Gregory Wornell 4/7/2026

Decocted Experience Improves Test-Time Inference in LLM Agents

Decocted Experience improves test-time inference for LLM agents by optimizing exploration budgets during reasoning and search without updating model parameters.

Ax Yanyuan Wang, Xiaowei Zhang 4/7/2026

Optimizing Service Operations via LLM-Powered Multi-Agent Simulation

LLM-powered multi-agent simulation framework for service operation optimization that models human behavior response to design choices via stochastic optimization.

Ax Jiayu Fu, Mourad Heddaya, Chenhao Tan 4/7/2026

Automatically Generating Hard Math Problems from Hypothesis-Driven Error Analysis

Automatically generates hard math problems from error analysis to evaluate LLM mathematical capabilities, addressing overfitting in math benchmarks.

Ax Yuwen Zhai, Runze Li, Liang Wang, Nian Shi, Liwu Xu, Wei Zhang, Ran Lin, Bo Xu, Benlei Cui 4/7/2026

GUIDE: Interpretable GUI Agent Evaluation via Hierarchical Diagnosis

GUIDE proposes hierarchical diagnosis framework for interpretable evaluation of GUI agents, providing diagnostic insights into long-horizon task failures beyond binary verdicts.

Ax Seohyeon Shin, HanJun Choi, Jun-Hyung Park, Hongkook Kim, Mansu Kim 4/7/2026

MolDA: Molecular Understanding and Generation via Large Language Diffusion Model

MolDA uses diffusion models instead of autoregressive generation for molecular understanding and generation, addressing sequential generation errors in LLM-based molecular discovery.

Ax Zhuowen Yuan, Zhaorun Chen, Zhen Xiang, Nathaniel D. Bastian, Seyyed Hadi Hashemi, Chaowei Xiao, Wenbo Guo, Bo Li 4/7/2026

ShieldNet: Network-Level Guardrails against Emerging Supply-Chain Injections in Agentic Systems

ShieldNet addresses supply-chain injection attacks in agentic systems where malicious behaviors are embedded in third-party tools and MCP servers used by LLM agents.

Ax Jihyun Lee, Yejin Min, Yejin Jeon, SungJun Yang, Hyounghun Kim, Gary Geunbae Lee 4/7/2026

PSY-STEP: Structuring Therapeutic Targets and Action Sequences for Proactive Counseling Dialogue Systems

Introduces STEP dataset and STEPPER model for cognitive behavioral therapy counseling agents that identify and address automatic negative thoughts in dialogue.

Ax Abu Noman Md Sakib, Zhensen Wang, Merjulah Roby, Zijie Zhang 4/7/2026

Empirical Characterization of Rationale Stability Under Controlled Perturbations for Explainable Pattern Recognition

Proposes metrics to assess consistency of attribution patterns in explainable AI systems under controlled input perturbations for reliable pattern recognition.

Ax Xiujiang Tan (Guangzhou Academy of Fine Arts, Guangzhou, China) 4/7/2026

The Topology of Multimodal Fusion: Why Current Architectures Fail at Creative Cognition

Analyzes structural limitations in multimodal AI architectures, arguing that contrastive alignment and cross-attention fusion fail at creative cognition due to modal separability assumptions.

Ax Jeonghwan Choi, Jibin Hwang, Gyeonghun Sun, Minjeong Ban, Taewon Yun, Hyeonjae Cheon, Hwanjun Song 4/7/2026

What Makes a Sale? Rethinking End-to-End Seller--Buyer Retail Dynamics with LLM Agents

RetailSim simulates end-to-end retail dynamics using LLM agents to evaluate seller persuasion, buyer-seller interaction, and purchase decisions across multiple stages.

Ax Jingyang Qiao, Weicheng Meng, Yu Cheng, Zhihang Lin, Zhizhong Zhang, Xin Tan, Jingyu Gong, Kun Shao, Yuan Xie 4/7/2026

Memory Intelligence Agent

Memory Intelligence Agent proposes a novel memory evolution system for deep research agents that improves trajectory retrieval and reduces storage/retrieval costs while enabling autonomous evolution.

Ax Varun Pratap Bhardwaj 4/7/2026

SuperLocalMemory V3.3: The Living Brain -- Biologically-Inspired Forgetting, Cognitive Quantization, and Multi-Channel Retrieval for Zero-LLM Agent Memory Systems

SuperLocalMemory V3.3 presents a local-first agent memory system with biologically-inspired forgetting, cognitive quantization, and multi-channel retrieval for AI coding agents without cloud LLM dependency.

Ax Daniele Foffano, Alessio Russo, Alexandre Proutiere 4/7/2026

Receding-Horizon Control via Drifting Models

Trajectory optimization approach using offline dataset and drifting learned models without requiring forward simulation in unknown dynamics settings.

Ax Hongju Pae 4/7/2026

Same World, Differently Given: History-Dependent Perceptual Reorganization in Artificial Agents

Architecture for artificial agents with history-dependent perceptual organization through feedback between slow perspective latents and perception processing.

Ax Yizhou Liu, Qi Sun, Yulin Chen, Siyue Zhang, Chen Zhao 4/7/2026

Search, Do not Guess: Teaching Small Language Models to Be Effective Search Agents

Methods for distilling search agent behaviors from LLMs into smaller language models for knowledge-intensive multi-hop reasoning tasks with lower computational costs.

Ax Seamus Brady 4/7/2026

Springdrift: An Auditable Persistent Runtime for LLM Agents with Case-Based Memory, Normative Safety, and Ambient Self-Perception

Persistent runtime system for long-lived LLM agents with auditable execution, case-based memory, deterministic safety gating, and self-perception capabilities.

Ax Nima H. Siboni 4/7/2026

On the "Causality" Step in Policy Gradient Derivations: A Pedagogical Reconciliation of Full Return and Reward-to-Go

Pedagogical clarification of the causality step in policy gradient derivations, explaining reward-to-go substitution in REINFORCE estimators with explicit mathematical rigor.

Ax Eranga Bandara, Asanga Gunaratna, Ross Gore, Abdul Rahman, Ravi Mukkamala, Sachin Shetty, Sachini Rajapakse, Isurunima Kularathna, Peter Foytik, Safdar H. Bouk, Xueping Liang, Amin Hass, Ng Wee Keong, Kasun De Zoysa 4/7/2026

AI Trust OS -- A Continuous Governance Framework for Autonomous AI Observability and Zero-Trust Compliance in Enterprise Environments

Framework for continuous governance, observability, and compliance of LLM, RAG, and multi-agent AI systems in enterprise environments using zero-trust principles.

Ax Xu Mingze 4/7/2026

ANX: Protocol-First Design for AI Agent Interaction with a Supporting 3EX Decoupled Architecture

ANX protocol-first framework for AI agent interaction with decoupled 3EX architecture, addressing token consumption, security, and fragmentation in agent-native protocols.

Ax Shu Wang, Edwin Yu, Oscar Love, Tom Zhang, Tom Wong, Steve Scargall, Charles Fan 4/7/2026

MemMachine: A Ground-Truth-Preserving Memory System for Personalized AI Agents

MemMachine open-source memory system for LLM agents integrating short-term, episodic, and profile memory with ground-truth preservation across multi-session interactions.

Ax Munawar Hasan 4/7/2026

Incompleteness of AI Safety Verification via Kolmogorov Complexity

Proves information-theoretic limits of AI safety verification using Kolmogorov complexity, showing incompleteness arises from fundamental constraints beyond computational complexity.

Ax Alexis Burgon, Berkman Sahiner, Nicholas A Petrick, Gene Pennello, Ravi K Samala 4/7/2026

Learning, Potential, and Retention: An Approach for Evaluating Adaptive AI-Enabled Medical Devices

Novel evaluation approach for adaptive AI medical devices using learning, potential, and retention metrics to assess iterative model updates and dataset changes.

Ax LM-Provers, Yuxiao Qu, Amrith Setlur, Jasper Dekoninck, Edward Beeching, Jia Li, Ian Wu, Lewis Tunstall, Aviral Kumar 4/7/2026

QED-Nano: Teaching a Tiny Model to Prove Hard Theorems

QED-Nano trains small models for mathematical theorem proving, achieving performance on complex proofs while remaining reproducible and efficient compared to proprietary systems.

Ax Ummara Mumtaz, Awais Ahmed, Summaya Mumtaz 4/7/2026

LLMs-Healthcare : Current Applications and Challenges of Large Language Models in various Medical Specialties

Comprehensive review of LLM applications in healthcare across medical specialties including diagnosis, treatment planning, and patient support with challenges analysis.

Ax Junhao Ye, Yuchen Hu, Ke Xu, Dingrong Pan, Qichun Chen, Jie Zhou, Shuai Zhao, Xinwei Fang, Xi Wang, Nan Guan, Zhe Jiang 4/7/2026

From Concept to Practice: an Automated LLM-aided UVM Machine for RTL Verification

Automated LLM-aided system for Universal Verification Methodology testbench construction and stimulus generation, reducing manual IC verification effort.

Ax Ruth Cohen, Lu Feng, Ayala Bloch, Sarit Kraus 4/7/2026

The Persuasion Paradox: When LLM Explanations Fail to Improve Human-AI Team Performance

Identifies Persuasion Paradox: fluent LLM explanations increase user confidence without improving task accuracy and sometimes undermining performance in human-AI teams.

Ax Xun Sun, Baiheng Xie, Li Huang, Qiang Gao 4/7/2026

Scaling DPPs for RAG: Density Meets Diversity

Scales determinantal point processes for RAG to improve diversity in retrieved context, addressing redundancy in standard relevance-ranking retrieval pipelines.

Ax Lily Jiaxin Wan, Chia-Tung Ho, Yunsheng Bai, Cunxi Yu, Deming Chen, Haoxing Ren 4/7/2026

FVRuleLearner: Operator-Level Reasoning Tree (OP-Tree)-Based Rules Learning for Formal Verification

FVRuleLearner uses LLM reasoning and operator-level reasoning trees for formal verification, automating natural language to SystemVerilog assertion translation.

Ax Jocelyn Beauchesne, Christine Maroti, Jeshua Bratman, Jerome Pesenti, Laurence Holt, Alex Tambellini, Allison McGrath, Matthew Guo, Sarah Peterson 4/7/2026

Personalized AI Practice Replicates Learning Rate Regularity at Scale

Studies learning rate regularity across 1.8M student interactions on Campus AI platform, automatically generating knowledge components without manual cognitive modeling.

Ax Daniel Grimes, Rachel M. Harrison 4/7/2026

BLK-Assist: A Methodological Framework for Artist-Led Co-Creation with Generative AI Models

BLK-Assist framework for artist-led co-creation with generative AI using parameter-efficient fine-tuning methods including LoRA and LayerDiffuse for image generation.

Ax Lucas G. Uberti-Bona Marin, Bram Rijsbosch, Kristof Meding, Gerasimos Spanakis, Gijs van Dijck, Konrad Kollnig 4/7/2026

Is your AI Model Accurate Enough? The Difficult Choices Behind Rigorous AI Development and the EU AI Act

Challenges technical objectivity of AI accuracy measurement, showing evaluation depends on context-dependent normative choices affecting error prioritization and risk distribution.

Ax Minghe Shen, Ananth Balashankar, Adam Fisch, David Madras, Miguel Rodrigues 4/7/2026

Robust LLM Performance Certification via Constrained Maximum Likelihood Estimation

Proposes constrained maximum likelihood estimation approach for rigorous LLM failure rate estimation, addressing tradeoff between expensive gold standards and biased automatic annotation.

Ax Xinhao Huang, You-Liang Huang, Zeyi Wen 4/7/2026

SoLA: Leveraging Soft Activation Sparsity and Low-Rank Decomposition for Large Language Model Compression

SoLA presents training-free compression method for LLMs using soft activation sparsity and low-rank decomposition without special hardware or expensive post-training.

Ax Hengshuai Yao, Xing Chen, Ahmed Murtadha, Jin Li, Shuai Shao, Yasin Abbasi Yadkori, Guan Wang, Mingli Yuan, William Chen, Sen Song 4/7/2026

Why Attend to Everything? Focus is the Key

Focus method learns token pair importance through learnable centroids, enabling efficient attention with minimal trainable parameters and zero downstream degradation.

Ax Horatio Morgan 4/7/2026

AI Governance Control Stack for Operational Stability: Achieving Hardened Governance in AI Systems

AI Governance Control Stack introduces layered governance architecture for operational stability, reliability, and auditability in high-stakes AI deployments.

Ax Keqin Xie 4/7/2026

LPC-SM: Local Predictive Coding and Sparse Memory for Long-Context Language Modeling

LPC-SM proposes hybrid autoregressive architecture separating local attention, persistent memory, and predictive correction for long-context language modeling using Orthogonal Novelty Transport.

Ax Gerardo Valente Vazquez-Garcia, Carmina Perez Guerrero, Eduardo Gardu\~no, Miguel Gonzalez-Mendoza, Adriana Palacios, Gerardo Rodriguez-Hernandez, Vahid Foroughi, Alba \`Agueda, Elsa Pastor, Gilberto Ochoa-Ruiz 4/7/2026

A reconfigurable smart camera implementation for jet flames characterization based on an optimized segmentation model

Smart FPGA camera platform with optimized AI models for real-time jet flame detection and characterization in industrial fire safety.

Ax Chenghao Huang, Jiarong Fan, Weiqing Wang, Hao Wang 4/7/2026

Safe Decentralized Operation of EV Virtual Power Plant with Limited Network Visibility via Multi-Agent Reinforcement Learning

Multi-agent reinforcement learning for decentralized EV virtual power plant operation with limited network visibility.

Ax Xiaowen Ma, Onur Ayan, Yunpu Ma, Xueli An 4/7/2026