Isolater - Feed

Ax Robert Richardson, Josh Meyers, Brian Hartman, David Sandberg 19d ago

Agentic AI and Retrieval-Augmented Models in Straight-Through Underwriting

Agentic AI systems for underwriting that combine RAG and multi-agent workflows to reason over unstructured documents in regulated decision workflows.

Ax Benjamin Poole, Minwoo Lee 19d ago

Feedback Manipulation Regularization: Enabling Offline Agent Alignment for Imitation Learning

Feedback Manipulation Regularization enables offline agent alignment using human demonstrations and feedback for imitation learning in RL agents.

Ax Tingkai Liu, Muralidhar Andoorveedu, Sanjoy Das, Sanjay Patel, Volodymyr Kindratenko 19d ago

CTA-Pipelining: A Latency-Oriented Spatial Scaling Method for Multi-GPU Systems

CTA-Pipelining optimizes multi-GPU systems for LLM serving latency by treating GPU interconnects as shared memory rather than networks.

Ax Felix Wang, Anudeep Das, Mei Nagappan, N. Asokan 19d ago

Functional and Secure Code Generation with Task Vectors

Task vectors approach for improving functional and secure code generation in LLMs by jointly optimizing correctness and vulnerability prevention.

Ax Luke Baines, Anton Gonzalvez Hawthorne, Mariia Koroliuk, Irakli Shalibashvili, Cl\'ement Dumas, Konstantinos Voudouris, David Demitri Africa 19d ago

Persona Cartography: Charting Language Model Personality Traits in Weight Space

Method for decomposing and controlling LLM personas using low-rank adapters and OCEAN personality trait framework.

Ax Wenqi Huang, Charley Lee, Leonard Tng, Serena Ge 19d ago

DeepSWE: Measuring Frontier Coding Agents on Original, Long-Horizon Engineering Tasks

DeepSWE: Benchmark of 113 original long-horizon software engineering tasks for evaluating AI coding agents on novel problems.

Ax Viet Vu, Renyuan Xu, Jiacheng Zhang, Yufei Zhang 19d ago

Expressivity and Statistical Trade-offs in Diffusion Policy Learning

Theoretical analysis of expressivity and statistical trade-offs in diffusion-based policy learning for reinforcement learning.

Ax Xiuyi Lou, Zicheng Xu, Yu-Neng Chuang, Hoang Anh Duy Le, Zhaozhuo Xu, Guanchu Wang, Vladimir Braverman 19d ago

When Implausible Tokens Get Reinforced: Tail-Aware Credit Calibration for LLM Reinforcement Learning

Tail-aware credit calibration for LLM reinforcement learning addressing positive-credit contamination of low-probability tokens.

Ax Yufei Xia, Anjun Gao, Yueyang Quan, Zhuqing Liu, Minghong Fang 19d ago

Who Broke the System? Failure Localization in LLM-Based Multi-Agent Systems

Framework for localizing failures in multi-agent LLM systems by identifying which agent caused system-level failures.

Ax Shiping Yang, Shining Liang, Weihao Liu, Wenbiao Ding, Linjun Shou, Lu Cheng, Angel X. Chang 19d ago

Hallucination Self-Play: Bootstrapping Reinforced Detector via Evolved Generator

Hallucination Self-Play: Method for bootstrapping hallucination detectors through iterative generator-detector co-evolution without static data.

Ax Kalle Kujanp\"a\"a, Ning Liu, Shahnawaz Alam, Yeshwanth Reddy Sura, Tianyu Yang, Kristina Klinkner, Shervin Malmasi 19d ago

Tool-Making and Self-Evolving LLM Agents in Low-Latency Systems

Production LLM agents with tool-making pipeline that compiles repeated procedural steps into validated tools to reduce inference latency.

Ax Anjun Gao, Yueyang Quan, Zhuqing Liu, Minghong Fang 19d ago

Beware What You Autocomplete: Forensic Attribution of Backdoored Code Completions

CodeTracer: Forensic framework for detecting and attributing backdoor attacks in LLM-based code completion systems.

Ax Emily Jin, Joy Hsu, Yiqing Xu, Weiyu Liu, Nick Haber, Jiajun Wu 19d ago

APIVOT: Adaptive Planning with Interleaved Vision-Language Thoughts

APIVOT: Vision-language model-based robot planner that interleaves language and visual reasoning for long-horizon task planning.

Ax Jennifer Za, Julija Bainiaksina, Nikita Ostrovsky, Tanush Chopra, Victoria Krakovna 19d ago

Persuasion Attacks Can Decrease Effectiveness of CoT Monitoring

Study showing chain-of-thought monitoring in AI agents can be bypassed through persuasion attacks and adversarial prompts.

Ax Andrej Leban, Yuekai Sun 19d ago

CausalDS: Benchmarking Causal Reasoning in Data-Science Agents

Benchmark for evaluating causal reasoning capabilities in data-science agents combining LLMs with tool use.

Ax Jun Nie, Yonggang Zhang, Jun Song, Qianshu Cai, Dahai Yu, Yike Guo, Xinmei Tian, Bo Han 19d ago

TTHE: Test-Time Harness Evolution

Test-time optimization framework evolving LLM agent harnesses to adapt executable programs for changing distributions.

Ax Zifan Zhang, Minghong Fang, Dianwei Chen, Zhuqing Liu, Prashant Khanduri, Xianfeng Yang, Anupam Das, Yuchen Liu 19d ago

Securing Autonomous Vehicle Systems via Twin-Aware Federated Reinforcement Learning

Federated reinforcement learning approach for autonomous vehicles resilient to poisoning attacks.

Ax Hongkang Yang, Zhi-Qin John Xu, Feiyu Xiong, Weinan E 19d ago

A First-Principles Theory of Slow Thinking and Active Perception

Mathematical theory formalizing slow thinking and active perception in large language models.

Ax Abdullah Al Shafi, Sumaiya Rahim Suma 19d ago

Closing the Null Space: Guidance-Aware Quantization for Classifier-Free Diffusion

Quantization method for classifier-free diffusion models accounting for paired conditional/unconditional structure.

Ax Angus Phillips, Gavin Kerrigan, Tom Rainforth 19d ago

Bayesian Experimental Design via Score Matching

Policy-based Bayesian experimental design using score matching to address expected information gain tractability.

Ax Thibaud Ardoin, Semira Einsele, Evis Bregu, Gerhard Wunder 19d ago

Prompt Compression via Activation Aggregation

Method for compressing LLM prompts into single activation vectors extracted from intermediate layers and re-injected earlier, reducing token overhead.

Ax Zheng Gao, Xiaoyu Li, Xiaoyan Feng, Jiaojiao Jiang, Yang Song, Yulei Sui, Zhenchang Xing, Liming Zhu 19d ago

TRACE: A Two-Channel Robust Attribution Watermark via Complementary Embeddings for LLM-Agent Trajectories

TRACE: Watermarking technique for LLM-agent trajectories using complementary embeddings to verify agent provenance and prevent model substitution.

Ax Ali Motahharynia, Mohammadreza Ghaffarzadeh-Esfahani, Mahsa Sheikholeslami, Navid Mazrouei, Matin Irajpour, Yousof Gheisari, Hajar Sirous 19d ago

DrugGen 2: A disease-aware language model for enhancing drug discovery

DrugGen-2: Language model for drug discovery conditioned on disease ontology and target protein sequences, fine-tuned for molecular generation.

Ax Jiawei Liang, Haotong Qin, Linfeng Du, Xingyu Liu, Shangkun Li, Hui Yu, Michele Magno, Xinyu Chen, Jiang Xu, Wei Zhang 19d ago

FPGN: Redefining Ultra-Fast Programmable Gate-based Neural Acceleration with Differentiable LUTs

FPGA-based neural accelerator using differentiable LUTs for nanosecond-scale DNN inference with ultra-low latency.

Ax Zijie Cheng, Yang Peng, Zhihua Zhang 19d ago

Statistical Efficiency and Inference of Quantile Distributional Reinforcement Learning

Statistical efficiency analysis of quantile-based distributional reinforcement learning for policy evaluation and return distribution characterization.

Ax Feng Wang, Canmiao Fu, Zhipeng Huang, Chen Li, Jing Lyu, Ge Li 19d ago

Cognitive-structured Multimodal Agent for Multimodal Understanding, Generation, and Editing

Cognitive-structured multimodal agent with episodic memory for long-horizon multimodal dialogue, understanding, generation, and editing tasks.

Ax Weiduo Liao, Yunqiao Yang, Ying Wei 19d ago

When Structured Sparse Autoencoders Learn Consistent Concepts Across Modalities

Structured sparse autoencoders for learning consistent concepts across modalities in vision-language models with improved interpretability.

Ax Kaifeng Zhao, Mathis Petrovich, Haotian Zhang, Tingwu Wang, Siyu Tang, Davis Rempe 19d ago

ARDY: Autoregressive Diffusion with Hybrid Representation for Interactive Human Motion Generation

Autoregressive diffusion model for real-time 3D human motion generation with text and kinematic constraints in interactive applications.

Ax Sameera Ramasinghe, Thalaiyasingam Ajanthan, Gil Avraham, Yan Zuo, Alexander Long 19d ago

Subspace Networks: Scaling Decentralized Training with Communication-Efficient Model Parallelism

Subspace Networks enable decentralized training of large models with communication-efficient model parallelism.

Ax Rapha\"el Bayle, Martial Mermillod, Robert M. French 19d ago

Are Current Continual Learning Methods Truly Agnostic? Introducing OPRE, a Step Toward Agnostic Continual Learning

OPRE method addresses catastrophic forgetting in continual learning by introducing truly agnostic evaluation approaches.

Ax Yuqi Yang, Ying Jin 19d ago

Multi-Distribution Robust Conformal Prediction

Conformal prediction method that provides uniform coverage guarantees across multiple source distributions for robust fairness.

Ax Xingyue Huang, Xueying Ding, Mingxuan Ju, Yozen Liu, Neil Shah, Tong Zhao 19d ago

Threshold Differential Attention for Sink-Free, Ultra-Sparse, and Non-Dispersive Language Modeling

Threshold Differential Attention mechanism addressing softmax limitations for long contexts with ultra-sparse, sink-free attention.

Ax Alexandre Myara, Nicolas Bourriez, Thomas Boyer, Thomas Lemercier, Ihab Bendidi, Auguste Genovesio 19d ago

XFACTORS: Disentangled Information Bottleneck via Contrastive Supervision

XFACTORS: disentangled representation learning via information bottleneck with contrastive supervision for semantic factor discovery.

Ax Weiqi Wang, Xin Liu, Binxuan Huang, Hejie Cui, Rongzhi Zhang, Changlong Yu, Shuowei Jin, Jingfeng Yang, Qingyu Yin, Zhengyang Wang, Zheng Li, Yifan Gao, Priyanka Nigam, Bing Yin, Lihong Li, Yangqiu Song 19d ago

HeaPA: Difficulty-Aware Heap Sampling and On-Policy Query Augmentation for LLM Reinforcement Learning

HeaPA: difficulty-aware heap sampling and on-policy query augmentation for efficient LLM reinforcement learning on reasoning tasks.

Ax Moritz Miller, Florent Draye, Bernhard Sch\"olkopf 19d ago

Towards Isolated Interventions via Almost Orthogonal Features in Language Models

Mechanistic interpretability approach using almost-orthogonal features in language models to enable isolated, reliable interventions.

Ax Jan Malte T\"opperwien, Aditya Mohan, Marius Lindauer 19d ago

Beyond Success Rates: Trainability and Extractability for Offline GCRL

Analysis of offline goal-conditioned reinforcement learning methods beyond success rates, measuring trainability and policy extractability.

Ax Bowen Yu, Maolin Wang, Sheng Zhang, Binhao Wang, Yi Wen, Jingtong Gao, Bowen Liu, Zimo Zhao, Wanyu Wang, Xiangyu Zhao 19d ago

Curriculum Learning for Efficient Chain-of-Thought Distillation via Structure-Aware Masking and GRPO

Curriculum learning framework for distilling chain-of-thought reasoning from large language models into smaller student models via structure-aware masking.

Ax Davide Tugnoli, Andrea De Lorenzo, Marco Virgolin, Giovanni Cin\`a 19d ago

Improving TabPFN's Synthetic Data Generation by Integrating Causal Structure

TabPFN foundation model improved with causal structure integration for better synthetic tabular data generation.

Ax Mumuksh Tayal, Manan Tayal, Ravi Prakash 19d ago

Safe Flow Q-Learning: Offline Safe Reinforcement Learning with Reachability-Based Flow Policies

Safe Flow Q-Learning algorithm for offline safe reinforcement learning combining reachability analysis with Q-learning for safety-critical control.

Ax Dibakar Sigdel 19d ago

The Phasor Transformer: Resolving Attention Bottlenecks on the Unit Circle

Phasor Transformer block replacing dot-product attention with phase-native computation on unit circle to address long-context sequence learning bottlenecks.

Ax Zhiyuan Chen, Yuxuan Zhong, Fan Wang, Bo Yu, Pengtao Shao, Shaoshan Liu, Ning Ding 19d ago

StateLinFormer: Stateful Training Enhancing Long-term Memory in Navigation

Research on stateful linear-attention navigation model enabling long-term memory across extended interactions.

Ax Iv\'an Olarte Rodr\'iguez, Anja Jankovic, Thomas B\"ack, Elena Raponi 19d ago

Does Dimensionality Reduction via Random Projections Preserve Landscape Features?

Research on whether random projections preserve landscape features in exploratory landscape analysis for optimization.

Ax Yikai Wang, Shang Liu, Jose Blanchet 19d ago

Wasserstein Distributionally Robust Regret Optimization for Reinforcement Learning from Human Feedback

Research on distributionally robust optimization for RLHF training to address reward misspecification and over-optimization in LLM alignment.

Ax Liang Luo, Yinbin Ma, Quanyu Zhu, Vasiliy Kuznetsov, Yuxin Chen, Neng Shi, Jian Jiao, Jiecao Yu, Buyun Zhang, Tongyi Tang, Xiaohan Wei, Yanli Zhao, Zeliang Chen, Yuchen Hao, Venkatesh Ranganathan, Sandeep Parab, Yantao Yao, Maxim Naumov, Chunzhi Yang, Shen Li, Ellie Wen, Wenlin Chen, Santanu Kolay, Chunqiang Tang 19d ago