Ax Apostol Vassilev 25d ago

Robust AI Security and Alignment: A Sisyphean Endeavor?

Information-theoretic analysis extending Gödel's incompleteness to AI security and alignment, establishing fundamental limitations for robust AI systems.

Ax Runze Li, Yuwen Zhai, Bo Xu, LiWu Xu, Nian Shi, Wei Zhang, Ran Lin, Liang Wang 25d ago

EchoTrail-GUI: Building Actionable Memory for GUI Agents via Critic-Guided Self-Exploration

Framework enabling GUI agents to build actionable memory from past tasks via self-exploration with critic guidance, improving generalization and reducing errors.

Ax Haoran Sun, Yongjian Guo, Zhong Guan, Shuai Di, Xiaodong Bai, Jing Long, Tianyun Zhao, Mingxi Luo, Hongke Zhao, Likang Wu, Xiaotie Deng, Xu Chu, Xi Xiao, Sheng Wen, Yicheng Gong, Junwu Xiong 25d ago

RL-VLA$^3$: A Flexible and Asynchronous Reinforcement Learning Framework for VLA Training

Asynchronous reinforcement learning framework for vision-language-action model training, enabling flexible post-training optimization for embodied agents.

Ax Harvey Lederman, Kyle Mahowald 25d ago

Emergent Introspection in AI is Content-Agnostic

Study demonstrating that introspection mechanisms in LLMs are content-agnostic, detecting anomalies without understanding their semantic meaning.

Ax Liang Ding 25d ago

AgentHER: Hindsight Experience Replay for LLM Agent Trajectory Relabeling

Framework adapting hindsight experience replay to recover training signal from failed LLM agent trajectories, addressing low real-world task success rates.

Ax Zining Fang, Cheng Xue, Chunhui Liu, Bin Xu, Ming Chen, Xiaowei Hu 25d ago

PhySe-RPO: Physics and Semantics Guided Relative Policy Optimization for Diffusion-Based Surgical Smoke Removal

Diffusion-based surgical video restoration framework using physics and semantics-guided reinforcement learning to remove surgical smoke.

Ax Difan Jiao, Qianfeng Wen, Blair Yang, Zhenwei Tang, Ashton Anderson 25d ago

ThinkTwice: Jointly Optimizing Large Language Models for Reasoning and Self-Refinement

Two-phase training framework jointly optimizing LLMs for reasoning and self-refinement using group relative policy optimization on correctness rewards.

Ax Xue Liu, Xin Ma, Yuxin Ma, Yongchang Peng, Duo Wang, Zhoufutu Wen, Ge Zhang, Kaiyuan Zhang, Xinyu Chen, Tianci He, Jiani Hou, Liang Hu, Ziyun Huang, Yongzhe Hui, Jianpeng Jiao, Chennan Ju, Yingru Kong, Yiran Li, Mengyun Liu, Luyao Ma, Fei Ni, Yiqing Ni, Yueyan Qiu, Yanle Ren, Zilin Shi, Zaiyuan Wang, Wenjie Yue, Shiyu Zhang, Xinyi Zhang, Kaiwen Zhao, Zhenwei Zhu, Shanshan Wu, Qi Zhao, Wenhao Huang 25d ago

Xpertbench: Expert Level Tasks with Rubrics-Based Evaluation

High-fidelity benchmark with rubrics-based evaluation assessing LLMs on expert-level complex open-ended tasks across multiple domains.

Ax Hang Xu, Ling Yue, Chaoqian Ouyang, Yuchen Liu, Libin Zheng, Shaowu Pan, Shimin Di, Min-Ling Zhang 25d ago

FactReview: Evidence-Grounded Reviews with Literature Positioning and Execution-Based Claim Verification

LLM-based peer review system that verifies claims by checking related work and executing code, improving review quality beyond manuscript-only analysis.

Ax Saad Alqithami 25d ago

Soft Tournament Equilibrium

Theoretical framework for evaluating cyclic non-transitive interactions between LLM-based agents using equilibrium concepts instead of linear rankings.

Ax Zhimin Zhao 25d ago

Gradual Cognitive Externalization: From Modeling Cognition to Constituting It

Framework proposing that ambient AI systems transition from modeling to constituting users' cognitive functions through sustained causal coupling.

Ax Seohyeon Shin, HanJun Choi, Jun-Hyung Park, Hong Kook Kim, Mansu Kim 25d ago

MolDA: Molecular Understanding and Generation via Large Language Diffusion Model

Molecular discovery framework combining LLMs with diffusion models to improve generation of chemically valid molecules by relaxing autoregressive constraints.

Ax Jingyang Qiao, Weicheng Meng, Yu Cheng, Zhihang Lin, Zhizhong Zhang, Xin Tan, Jingyu Gong, Kun Shao, Yuan Xie 25d ago

Memory Intelligence Agent

Memory system for deep research agents that improves trajectory retrieval and memory evolution to enhance LLM reasoning and autonomous learning.

Ax Md Zarif Hossain, Ahmed Imteaj 25d ago

Sim-CLIP: Unsupervised Siamese Adversarial Fine-Tuning for Robust and Semantically-Rich Vision-Language Models

Unsupervised fine-tuning method to improve adversarial robustness and semantic quality of vision-language models through siamese contrastive learning.

Ax Zhiqiang Yuan, Weitong Chen, Hanlin Wang, Xin Peng, Zhenpeng Chen, Yiling Lou 25d ago

TransAgent: Enhancing LLM-Based Code Translation via Fine-Grained Execution Alignment

LLM-based code translation agent using execution alignment to improve cross-language code generation without parallel training data.

Ax Fanrui Zhang, Jiawei Liu, Jiaying Zhu, Esther Sun, Dong Li, Qiang Zhang, Zheng-Jun Zha 25d ago

ForgeryGPT: A Multimodal LLM for Interpretable Image Forgery Detection and Localization

Multimodal LLM fine-tuned for image forgery detection and localization with interpretable visual reasoning capabilities.

Ax Saketh Ram Kasibatla, Arpan Agarwal, Yuriy Brun, Sorin Lerner, Talia Ringer, Emily First 25d ago

Cobblestone: A Divide-and-Conquer Approach for Automating Formal Verification

Divide-and-conquer proof synthesis approach using LLMs to automate formal verification in proof assistants like Coq, improving software quality verification.

Ax Gopi Krishnan Rajbahadur, Gustavo A. Oliva, Dayi Lin, Jiho Shin, Ahmed E. Hassan 25d ago

From Cool Demos to Production-Ready FMware: Core Challenges and a Technology Roadmap

Systematic analysis of challenges in transitioning foundation model systems from demos to production, covering reliability, cost, scalability, and compliance issues.

Ax Xiao Liu, Lijun Zhang, Deepak Ganesan, Hui Guan 25d ago

Aligned Vector Quantization for Edge-Cloud Collabrative Vision-Language Models

Edge-cloud collaborative VQA system using aligned vector quantization to split vision-language model computation between edge and cloud devices, reducing bandwidth and utilizing edge resources.

Ax Kutay Tire, Ege Onur Taga, Muhammed Emrullah Ildiz, Samet Oymak 25d ago

Retrieval Augmented Time Series Forecasting

Retrieval-augmented generation applied to time-series foundation models for zero-shot forecasting across domains.

Ax Junhyeok Kang, Yooju Shin, Jae-Gil Lee 25d ago

VarDrop: Enhancing Training Efficiency by Reducing Variate Redundancy in Periodic Time Series Forecasting

VarDrop reduces computational cost in multivariate time series forecasting by eliminating variate token redundancy.

Ax Hammad Ayyubi, Junzhang Liu, Ali Asgarov, Zaber Ibn Abdul Hakim, Najibul Haque Sarker, Zhecan Wang, Chia-Wei Tang, Hani Alomari, Md. Atabuzzaman, Xudong Lin, Naveen Reddy Dyava, Shih-Fu Chang, Chris Thomas 25d ago

ENTER: Event Based Interpretable Reasoning for VideoQA

ENTER system uses event graphs for interpretable Video QA with code generation and contextual reasoning.

Ax Hadi Zare, Mostafa Abbasi, Maryam Ahang, Homayoun Najjaran 25d ago

An Innovative Next Activity Prediction Using Process Entropy and Dynamic Attribute-Wise-Transformer in Predictive Business Process Monitoring

Entropy-based framework with Transformer for next activity prediction in business process monitoring.

Ax Penghui Yang, Cunxiao Du, Fengzhuo Zhang, Haonan Wang, Tianyu Pang, Chao Du, Bo An 25d ago

LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification

LongSpec enables efficient speculative decoding for long-context LLM inference with lossless acceleration for agent applications.

Ax Rafiya Javed, Cassandra Parent, Jackie Kay, David Yanni, Abdullah Zaini, Anushe Sheikh, Maribeth Rauh, Walter Gerych, Ramona Comanescu, Iason Gabriel, Marzyeh Ghassemi, Laura Weidinger 25d ago

Hedging and Non-Affirmation: Quantifying LLM Alignment on Questions of Human Rights

Framework measuring hedging and non-affirmation behaviors in LLM responses on human rights topics across identity groups.

Ax Firoj Alam, Md Arid Hasan, Sahinur Rahman Laskar, Mucahid Kutlu, Kareem Darwish, Shammur Absar Chowdhury 25d ago

NativQA Framework: Enabling LLMs and VLMs with Native, Local, and Everyday Knowledge

NativQA framework extends to multimodality for culturally-grounded LLM/VLM evaluation across languages and regions.

Ax Junhao Ye, Yuchen Hu, Ke Xu, Dingrong Pan, Qichun Chen, Jie Zhou, Shuai Zhao, Xinwei Fang, Xi Wang, Nan Guan, Zhe Jiang 25d ago

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

LLM-aided tool automates Universal Verification Methodology testbench generation for RTL IC verification.

Ax Darpan Aswal, Siddharth D Jaiswal 25d ago

Phonetic Perturbations Reveal Tokenizer-Rooted Safety Gaps in LLMs

CMP-RT diagnostic probe reveals tokenization vulnerabilities in safety-aligned LLMs through phonetic perturbations.

Ax Noah Amsel, David Persson, Christopher Musco, Robert M. Gower 25d ago

The Polar Express: Optimal Matrix Sign Methods and Their Application to the Muon Algorithm

Polar decomposition and matrix sign methods optimized for GPU-friendly deep learning training via Muon optimizer.

Ax Florian F\"urrutter, Zohim Chandani, Ikko Hamamura, Hans J. Briegel, Gorka Mu\~noz-Gil 25d ago

Synthesis of discrete-continuous quantum circuits with multimodal diffusion models

Multimodal diffusion models synthesize quantum circuits for efficient compilation with reduced hardware calls and runtimes.

Ax Yihan Xie, Sijing Li, Tianwei Lin, Zhuonan Wang, Chenglin Yang, Yu Zhong, Wenjie Yan, Wenqiao Zhang, Xiaogang Guo, Jun Xiao, Yueting Zhuang, Beng Chin Ooi 25d ago

HeartcareGPT: A Unified Multimodal ECG Suite for Dual Signal-Image Modeling and Understanding

HeartcareGPT suite with 400K ECG dataset enables multimodal medical LLMs for dual signal-image ECG understanding.

Ax Denis Rozumny, Jonathon Luiten, Numair Khan, Johannes Sch\"onberger, Peter Kontschieder 25d ago

BulletGen: Improving 4D Reconstruction with Bullet-Time Generation

BulletGen reconstructs 4D dynamic scenes from monocular video using generative models to complete unseen regions.

Ax Chaofan Pan, Xin Yang, Yanhua Li, Wei Wei, Tianrui Li, Bo An, Jiye Liang 25d ago

A Survey of Continual Reinforcement Learning

Survey of continual reinforcement learning covering sequential decision-making, generalization, and adaptation across dynamic tasks.

Ax Zihe Yan, Jiaping Gui, Zhuosheng Zhang, Gongshen Liu 25d ago

LaSM: Layer-wise Scaling Mechanism for Defending Pop-up Attack on GUI Agents

LaSM defends GUI agents against pop-up injection attacks using layer-wise scaling on multimodal LLMs for safer screen interaction.

Ax Joosung Lee, Cheonbok Park, Hwiyeol Jo, Jeonghoon Kim, Joonsuk Park, Kang Min Yoo 25d ago

Enhancing Hallucination Detection via Future Context

Framework for detecting LLM hallucinations in black-box generators by leveraging future context patterns.

Ax Phi Van Nguyen, Ngoc Huynh Trinh, Duy Minh Lam Nguyen, Phu Loc Nguyen, Quoc Long Tran 25d ago

Aleatoric Uncertainty Medical Image Segmentation Estimation via Flow Matching

Flow matching approach for quantifying aleatoric uncertainty in medical image segmentation, modeling expert annotation variability.

Ax Wangsong Yin, Daliang Xu, Mengwei Xu, Gang Huang, Xuanzhe Liu 25d ago

ShadowNPU: System and Algorithm Co-design for NPU-Centric On-Device LLM Inference

ShadowNPU enables efficient on-device LLM inference by redesigning attention operator for NPU execution, improving privacy and performance.

Ax Francisco Caetano, Christiaan Viviers, Peter H. N. De With, Fons van der Sommen 25d ago

MedShift: Implicit Conditional Transport for X-Ray Domain Adaptation

MedShift addresses domain gap between synthetic and real X-ray images using conditional transport for improved generalization to clinical settings.

Ax Junsong Li, Jie Zhou, Bihao Zhan, Yutao Yang, Qianjun Pan, Shilian Chen, Tianyu Huai, Xin Li, Qin Chen, Liang He 25d ago

LifeAlign: Lifelong Alignment for Large Language Models with Memory-Augmented Focalized Preference Optimization

LifeAlign framework for lifelong LLM alignment across sequential tasks using memory-augmented preference optimization without catastrophic forgetting.

Ax Ziyi Liu 25d ago

A State-Update Prompting Strategy for Efficient and Robust Multi-turn Dialogue

Training-free prompt engineering strategy using state reconstruction and history reminders for efficient multi-turn LLM dialogue.

Ax Suhas Suresh Bharadwaj, Prerana Ramkumar 25d ago

Chiplet-Based RISC-V SoC with Modular AI Acceleration

Chiplet-based RISC-V SoC architecture with modular AI acceleration for edge AI devices with improved yield and efficiency.

Ax Xingyu Shen, Yingfa Chen, Zhen Leng Thai, Xu Han, Zhiyuan Liu, Maosong Sun 25d ago

StateX: Enhancing RNN Recall via Post-training State Expansion

StateX post-training method improves recall ability in RNNs and state-space models for long-context information retrieval.

Ax Hemang Jain, Shailender Goyal, Divyansh Pandey, Karthik Vaidhyanathan 25d ago

Dissecting Transformers: A CLEAR Perspective towards Green AI

Component-level energy assessment framework analyzing transformer efficiency to enable green AI development.

Ax Noor Islam S. Mohammad, Md Muntaqim Meherab 25d ago

Reveal-to-Revise: Explainable Bias-Aware Generative Modeling with Multimodal Attention

Explainable bias-aware generative framework combining multimodal attention, attribution methods, and iterative feedback for fair generation.

Ax Junhoo Lee, Seungyeon Kim, Nojun Kwak 25d ago

Unlocking the Potential of Diffusion Language Models through Template Infilling

Template Infilling conditioning strategy enables diffusion language models to handle flexible structural prompting beyond prefix-based generation.

Ax Xingrui Zhuo, Jiapu Wang, Gongqing Wu, Zhongyuan Wang, Jichen Zhang, Shirui Pan, Xindong Wu 25d ago

Knowledge Reasoning Language Model: Unifying Knowledge and Language for Inductive Knowledge Graph Reasoning

Knowledge Reasoning Language Model unifies language models with knowledge graphs for inductive reasoning over unknown entities and relations.

Ax Qing Yang, Zhenghao Liu, Yangfan Du, Pengcheng Huang, Tong Xiao 25d ago

RLAIF-SPA: Structured AI Feedback for Semantic-Prosodic Alignment in Speech Synthesis

RLAIF-SPA uses structured AI feedback to improve emotional expressiveness and semantic-prosodic alignment in text-to-speech synthesis.

Ax Afrozah Nadeem, Mark Dras, Usman Naseem 25d ago

Fairness Evaluation and Inference Level Mitigation in LLMs

Methods for evaluating and mitigating fairness issues in LLMs at inference time to reduce harmful behaviors and drift.

Ax Youngjun Choi, Joonseong Kang, Sungjun Lim, Kyungwoo Song 25d ago

Eigen-Value: Efficient Domain-Robust Data Valuation via Eigenvalue-Based Approach

Eigen-Value method for efficient data valuation using eigenvalue-based approach, focusing on out-of-distribution robustness.

Ax Hao Huang, Geeta Chandra Raju Bethala, Shuaihang Yuan, Congcong Wen, Mengyu Wang, Anthony Tzes, Yi Fang 25d ago

One-shot Adaptation of Humanoid Whole-body Motion with Walking Priors

Data-efficient approach for adapting humanoid robot whole-body motion control from single motion examples using walking priors.