Ax Can Xie, Ruotong Pan, Xiangyu Wu, Yunfei Zhang, Jiayi Fu, Tingting Gao, Guorui Zhou 4/17/2026

Unlocking Exploration in RLVR: Uncertainty-aware Advantage Shaping for Deeper Reasoning

Uncertainty-aware advantage shaping method improving exploration in reinforcement learning for enhanced LLM reasoning capabilities.

Ax Rikard Rosenbacke, Carl Rosenbacke, Victor Rosenbacke, Martin McKee 4/17/2026

Beyond "Hallucinations": A Framework for Stable Human-AI Reasoning

Framework addressing LLM failures as cognitive rather than purely technical, distinguishing hallucinations from associative pattern reproduction.

Ax Wenhao Wang, Peizhi Niu, Zhao Xu, Zhaoyu Chen, Jian Du, Yaxin Du, Xianghe Pang, Keduan Huang, Yanfeng Wang, Qiang Yan, Siheng Chen 4/17/2026

MCP-Flow: Facilitating LLM Agents to Master Real-World, Diverse and Scaling MCP Tools

MCP-Flow framework enabling LLM agents to discover and utilize diverse Model Contextual Protocol tools at scale for real-world tasks.

Ax Ali Taheri, Alireza Taban, Sadegh Soudjani, Ashutosh Trivedi 4/17/2026

BarrierBench: Evaluating Large Language Models for Safety Verification in Dynamical Systems

BarrierBench evaluates LLMs on safety verification of dynamical systems using barrier certificates for autonomous applications.

Ax Tuan-Anh Le, Anh Mai Vu, David Yang, Akash Awasthi, Hien Van Nguyen 4/17/2026

IMACT-CXR: An Interactive Multi-Agent Conversational Tutoring System for Chest X-Ray Interpretation

IMACT-CXR multi-agent conversational tutoring system using AutoGen for interactive chest X-ray interpretation training.

Ax Xuyuan Liu, Shengyu Chen, Xinshuai Dong, Yanchi Liu, Xujiang Zhao, Haoyu Wang, Yujun Yan, Haifeng Chen, Zhengzhang Chen 4/17/2026

Representation Interventions Enable Lifelong Knowledge Memory Control in LLMs

RILKE method for efficient knowledge updates in LLMs via representation interventions without costly retraining in lifelong settings.

Ax Austin Spizzirri 4/17/2026

The Specification Trap: Why Static Value Alignment Alone Is Insufficient for Robust Alignment

Argues static value alignment insufficient for robust AI alignment under capability scaling and distributional shift; proposes dynamic approach.

Ax Haoyu Dong, Pengkun Zhang, Yan Gao, Xuanyu Dong, Yilin Cheng, Mingzhe Lu, Zikun Zhu, Adina Yakefu, Shuxin Zheng 4/17/2026

Finch: Benchmarking Finance & Accounting across Spreadsheet-Centric Enterprise Workflows

FinWorkBench benchmark evaluating AI agents on enterprise finance workflows spanning data entry, retrieval, calculation, and reporting tasks.

Ax Dongming Jiang, Yi Li, Guanpeng Li, Bingzhe Li 4/17/2026

MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents

MAGMA multi-graph memory architecture for agents separating temporal, causal, and entity information to improve retrieval and reasoning.

Ax Qirui Mi, Zhijian Ma, Mengyue Yang, Haoxuan Li, Yisen Wang, Haifeng Zhang, Jun Wang 4/17/2026

Skill-Pro: Learning Reusable Skills from Experience via Non-Parametric PPO for LLM Agents

Skill-Pro framework enabling LLM agents to learn and reuse procedural skills from experiences without parameter updates via non-parametric PPO.

Ax Siyuan Li, Yunjia Wu, Yiyong Xiao, Pingyang Huang, Peize Li, Ruitong Liu, Yan Wen, Te Sun 4/17/2026

Evolving Beyond Snapshots: Harmonizing Structure and Sequence via Entity State Tuning for Temporal Knowledge Graph Forecasting

Entity State Tuning method for temporal knowledge graph forecasting using stateful entity representations across timestamps.

Ax Tan Bui-Thanh 4/17/2026

The AI Research Assistant: Promise, Peril, and a Proof of Concept

Case study of human-AI collaboration discovering novel error bounds for Hermite quadrature rules in mathematical research.

Ax Drew Prinster, Clara Fannjiang, Ji Won Park, Kyunghyun Cho, Anqi Liu, Suchi Saria, Samuel Stanton 4/17/2026

Conformal Policy Control

Conformal inference method to regulate untested policies against safe reference policies for high-stakes agent deployment.

Ax Jiangyu Chen 4/17/2026

AI4S-SDS: A Neuro-Symbolic Solvent Design System via Sparse MCTS and Differentiable Physics Alignment

Neuro-symbolic solvent design system using sparse MCTS and differentiable physics to overcome LLM agent limitations in combinatorial chemistry.

Ax Alvin Rajkomar, Pavan Sudarshan, Angela Lai, Lily Peng 4/17/2026

The Validity Gap in Health AI Evaluation: A Cross-Sectional Analysis of Benchmark Composition

Analyzes composition gaps in health AI benchmarks validating clinical LLMs without defined patient population characteristics.

Ax Richard Servajean, Philippe Servajean 4/17/2026

Measuring the metacognition of AI

Develops methods to measure metacognitive capabilities of AI systems for uncertainty assessment and decision reliability.

Ax Hyunji Nam, Dorottya Demszky 4/17/2026

Mitigating LLM biases toward spurious social contexts using direct preference optimization

Uses direct preference optimization to reduce LLM biases toward spurious social contexts in high-stakes decision-making tasks.

Ax Albert Sadowski, Jaros{\l}aw A. Chudziak 4/17/2026

Rashomon Memory: Towards Argumentation-Driven Retrieval for Multi-Perspective Agent Memory

Proposes Rashomon Memory architecture for AI agents to maintain multiple conflicting interpretations of events for concurrent goals.

Ax Myeongsoo Kim, Joe Hsu, Dingmin Wang, Shweta Garg, Varun Kumar, Murali Krishna Ramanathan 4/17/2026

CODESTRUCT: Code Agents over Structured Action Spaces

CODESTRUCT reframes code editing from text manipulation to structured AST operations, enabling LLM code agents to apply syntax-validated transformations on repositories.

Ax Shantam Srivastava, Mahesh Bhosale, David Doermann, Mingchen Gao 4/17/2026

CWCD: Category-Wise Contrastive Decoding for Structured Medical Report Generation

CWCD applies category-wise contrastive decoding to multi-modal LLMs for improved chest X-ray medical report generation with enhanced diagnostic accuracy.

Ax Chen Huang, Zitan Jiang, Changyi Zou, Wenqiang Lei, See-Kiong Ng 4/17/2026

Towards Proactive Information Probing: Customer Service Chatbots Harvesting Value from Conversation

Introduces Proactive Information Probing task for customer service chatbots to strategically extract business intelligence while minimizing conversation friction and user effort.

Ax Charafeddine Mouzouni 4/17/2026

Context Kubernetes: Declarative Orchestration of Enterprise Knowledge for Agentic AI Systems

Context Kubernetes orchestrates enterprise knowledge delivery to agentic AI systems with proper permissions and freshness, formalizing six design dimensions for knowledge orchestration.

Ax Georgianna "Blue" Lin, Rencong Jiang, No\'emie Elhadad, Xuhai "Orson" Xu 4/17/2026

A longitudinal health agent framework

Proposes longitudinal health agent framework addressing user intent and accountability for multi-turn health tasks like symptom management and behavior change.

Ax Weikang Zhang, Zimo Zhu, Zhichuan Yang, Chen Huang, Wenqiang Lei, See-Kiong Ng 4/17/2026

Beyond Prompt: Fine-grained Simulation of Cognitively Impaired Standardized Patients via Stochastic Steering

StsPatient framework simulates cognitively impaired standardized patients for clinical training using fine-grained LLM steering across domain-specific deficits and severity levels.

Ax Lei Lin, Jizhao Zhu, Yong Liu, Donghong Sun, Hongbo He, Yihua Du 4/17/2026

Heuristic Classification of Thoughts Prompting (HCoT): Integrating Expert System Heuristics for Structured Reasoning into Large Language Models

Paper proposes HCoT, integrating expert system heuristics into LLMs for structured reasoning to address stochastic token generation and decoupled decision-making in complex problem-solving.

Ax Jintao Xue, Xiao Li, Nianmin Zhang 4/17/2026

Safe reinforcement learning with online filtering for fatigue-predictive human-robot task planning and allocation in production

Safe reinforcement learning with online filtering for human-robot task allocation in manufacturing that constrains worker fatigue while maximizing efficiency.

Ax Jintao Xue, Xiao Li, Nianmin Zhang 4/17/2026

A hierarchical spatial-aware algorithm with efficient reinforcement learning for human-robot task planning and allocation in production

Hierarchical spatial-aware reinforcement learning algorithm for human-robot task planning and allocation in manufacturing with efficient coordination.

Ax Serdar Kadioglu, Karthik Uppuluri, Akash Singirikonda 4/17/2026

Modeling Copilots for Text-to-Model Translation

Text2Model suite introduces LLM-based copilots for text-to-optimization translation with varying complexity strategies and cross-domain dataset Text2Zinc.

Ax Stefan Kambiz Behfar, Jon Crowcroft 4/17/2026

Towards Adaptive, Learning-Based Security in Decentralized Applications

Proposes learning-based adaptive security mechanisms for Web3 decentralized applications addressing social, application, and protocol-layer attacks.

Ax Martin Kuo, Jianyi Zhang, Dongting Li, Yiran Chen 4/17/2026

DA-Cramming: Enhancing Cost-Effective Language Model Pretraining with Dependency Agreement Integration

DA-Cramming improves cost-effective BERT pretraining by integrating dependency agreement, extending Cramming approach for single GPU training.

Ax Bo Shu, Yiting Zhang, Saisai Hu, Dong Shu 4/17/2026

Generative Models and Connected and Automated Vehicles: A Survey in Exploring the Intersection of Transportation and AI

Survey examining application of generative models in connected and automated vehicles for predictive modeling, simulation, and decision-making.

Ax Samiha Sharlin, Tyler R. Josephson 4/17/2026

In Context Learning and Reasoning for Symbolic Regression with Large Language Models

Explores LLM capabilities for symbolic regression using in-context learning with GPT-4 models to suggest expressions that are optimized against datasets.

Ax Yuwei Yin, Giuseppe Carenini 4/17/2026

Improving Language Models with Intentional Analysis

Introduces intentional analysis framework to improve language model reasoning by explicitly modeling intent as cognitive notion underlying human communication and problem-solving.

Ax Ido Galil, Moshe Kimhi, Ran El-Yaniv 4/17/2026

Maximal Brain Damage Without Data or Optimization: Disrupting Neural Networks via Sign-Bit Flips

Deep Neural Lesion (DNL) identifies critical parameters in DNNs through data-free and optimization-free methods showing vulnerability to single bit-flip attacks.

Ax Andrea Apicella, Salvatore Giugliano, Francesco Isgr\`o, Andrea Pollastro, Roberto Prevete 4/17/2026

IMPACTX: improving model performance by appropriately constraining the training with teacher explanations

IMPACTX leverages explainable AI as automated attention mechanism to improve model performance without external knowledge or manual constraint specification.

Ax Hiba Arnaout, Noy Sternlicht, Tom Hope, Iryna Gurevych 4/17/2026

In-depth Research Impact Summarization through Fine-Grained Temporal Citation Analysis

Generates nuanced, time-aware impact summaries of scientific papers by analyzing confirmation and correction citations across temporal evolution.

Ax Yixian Xu, Shengjie Luo, Liwei Wang, Di He, Chang Liu 4/17/2026

Diagnosing and Improving Diffusion Models by Estimating the Optimal Loss Value

Method for estimating optimal loss values in diffusion models to improve diagnosis and performance by distinguishing between large optimal loss and insufficient model capacity.

Ax Fardin Saad, Pradeep K. Murukannaiah, Munindar P. Singh 4/17/2026

Theory of Mind in Action: The Instruction Inference Task in Dynamic Human-Agent Collaboration

Proposes instruction inference task for human-agent collaboration where agents infer incomplete instructions by modeling human mental states and shared context understanding.

Ax Yiyuan Yang, Zichuan Liu, Lei Song, Kai Ying, Zhiguang Wang, Tom Bamford, Svitlana Vyetrenko, Jiang Bian, Qingsong Wen 4/17/2026

Time-RA: Towards Time Series Reasoning for Anomaly Diagnosis with LLM Feedback

Time-RA reformulates time series anomaly detection as a generative reasoning task using LLM feedback, introducing RATs40K dataset for fine-grained anomaly categorization and explanation.

Ax Xinwei Wu, Haojie Li, Hongyu Liu, Xinyu Ji, Ruohan Li, Yule Chen, Yigeng Zhang 4/17/2026

Uncovering the Fragility of Trustworthy LLMs through Chinese Textual Ambiguity

Study evaluates trustworthiness of LLMs on ambiguous Chinese text using a benchmark dataset of disambiguated sentence pairs to assess model behavior under linguistic ambiguity.

Ax Dai Do, Manh Nguyen, Svetha Venkatesh, Hung Le 4/17/2026

SPaCe: Unlocking Sample-Efficient Large Language Models Training With Self-Pace Curriculum Learning

SPaCe introduces self-paced curriculum learning to reduce data and compute requirements for LLM fine-tuning with reinforcement learning by sampling training examples based on difficulty and learning value.

Ax Yubo Gao, Renbo Tu, Gennady Pekhimenko, Nandita Vijaykumar 4/17/2026

DPQuant: Efficient and Differentially-Private Model Training via Dynamic Quantization Scheduling

DPQuant combines dynamic quantization scheduling with differentially-private SGD to reduce training time, energy, and cost while maintaining privacy.

Ax Ivan Ternovtsii, Yurii Bilak 4/17/2026

Cosine-Similarity Routing with Semantic Anchors for Interpretable Mixture-of-Experts Language Models

Semantic Resonance Architecture routes tokens to experts in MoE models via cosine similarity to learnable semantic anchors, improving interpretability of routing decisions.

Ax Aleksi Huotala, Miikka Kuutila, Olli-Pekka Turtio, Simo Sipil\"a, Mika M\"antyl\"a 4/17/2026

AISysRev -- LLM-based Tool for Title-abstract Screening

AISysRev is a containerized LLM-based tool for automated title-abstract screening in systematic literature reviews.

Ax Shangqing Tu, Yaxuan Li, Yushi Bai, Lei Hou, Juanzi Li 4/17/2026

DeepPrune: Parallel Scaling without Inter-trace Redundancy

DeepPrune optimizes parallel LLM reasoning by eliminating inter-trace redundancy, reducing 80% computational waste in multi-trace inference.

Ax Shinwoo Park, Hyejin Park, Hyeseon An, Yo-Sub Han 4/17/2026

A Linguistics-Aware LLM Watermarking via Syntactic Predictability

LLM watermarking technique using syntactic predictability for verifiable output attribution while preserving text quality.

Ax Jingyao Liu, Chen Huang, Zhizhao Guan, Wenqiang Lei, Yang Deng 4/17/2026

E2Edev: Benchmarking Large Language Models in End-to-End Software Development Task

E2EDev benchmark evaluates LLMs on end-to-end software development tasks with fine-grained requirements and behavior-driven evaluation.

Ax Wenshuo Wang, Ziyou Jiang, Junjie Wang, Mingyang Li, Jie Huang, Yuekai Huang, Zhiyuan Chang, Feiyan Duan, Qing Wang 4/17/2026

Fall into a Pit, Gain in a Wit: Cognitive-Guided Harmful Meme Detection via Misjudgment Risk Pattern Retrieval

PatMD detects harmful memes using MLLMs and misjudgment risk patterns. ML research on content moderation.

Ax Jiahao Tang, Henry Hengyuan Zhao, Lijian Wu, Yifei Tao, Dongxing Mao, Yang Wan, Jingru Tan, Min Zeng, Min Li, Alex Jinpeng Wang 4/17/2026

From Charts to Code: A Hierarchical Benchmark for Multimodal Models

Chart2Code benchmark evaluating chart understanding and code generation in multimodal LLMs across three difficulty levels with real-world scenarios.

Ax Zirui Deng, Netanel Raviv 4/17/2026

Efficient Vector Symbolic Architectures from Histogram Recovery

Vector symbolic architectures using histogram recovery from random linear codes for neurosymbolic AI systems and hardware implementations.