Ax Marc Dymetman 5/5/2026

Binary Rewards and Reinforcement Learning: Fundamental Challenges

Analysis of diversity collapse in RLVR training with binary rewards and fundamental structural limitations.

Ax Michael Helcig, Eldar Kurtic, Dan Alistarh 5/5/2026

Statistically-Lossless Quantization of Large Language Models

Statistically-lossless quantization framework for LLMs balancing compression with inference acceleration.

Ax Sofianos Panagiotis Fotias, Vassilis Gaganis 5/5/2026

Closed-Loop CO2 Storage Control With History-Based Reinforcement Learning and Latent Model-Based Adaptation

Deep reinforcement learning controllers for CO2 geological storage with history-based adaptation under partial observability.

Ax Mehryar Mohri, Jon Schneider, Yutao Zhong 5/5/2026

Generalized Distributional Alignment Games for Unbiased Answer-Level Fine-Tuning

Framework for unbiased answer-level fine-tuning using distributional alignment games with Bregman divergence.

Ax Yao Shu, Chenxing Wei, Hongbin Lin, Shuang Qiu, Hui Xiong 5/5/2026

Reference-Sampled Boltzmann Projection for KL-Regularized RLVR: Target-Matched Weighted SFT, Finite One-Shot Gaps, and Policy Mirror Descent

Reinforcement learning with verifiable rewards using Boltzmann projection and weighted SFT for policy optimization.

Ax Chenye Yang, Weiyu Xu, Lifeng Lai 5/5/2026

Efficient Preference Poisoning Attack on Offline RLHF

Study of preference poisoning attacks against offline RLHF/DPO training pipelines for language models.

Ax Joern Hentsch 5/5/2026

MPCS: Neuroplastic Continual Learning via Multi-Component Plasticity and Topology-Aware EWC

Neuroplastic continual learning architecture integrating 11 mechanisms including task-driven neurogenesis and EWC regularization.

Ax Firas Mohamed Elamine Kiram, Imane Youkana, Rachida Saouli, Gian Antonio Susto, Laid Kahloul 5/5/2026

Recurrent Deep Reinforcement Learning for Chemotherapy Control under Partial Observability

Recurrent deep RL with memory-augmented policies for optimizing chemotherapy dosing under partial observability.

Ax Mohamadsadegh Khosravani, Sandra Zilles 5/5/2026

Gradient-Discrepancy Acquisition for Pool-Based Active Learning

Novel gradient-based acquisition criterion for pool-based active learning derived from generalization bounds.

Ax Inoussa Mouiche 5/5/2026

Gradient-Gated DPO: Stabilizing Preference Optimization in Language Models

Method to stabilize Direct Preference Optimization in LLMs by addressing gradient squeezing effects in human feedback alignment.

Ax Xin Zhang, Qiqi Tao, Jiawei Du, Moyun Liu, Joey Tianyi Zhou 5/5/2026

Visual Latents Know More Than They Say: Unsilencing Latent Reasoning in MLLMs

Research on improving multimodal LLM latent reasoning by addressing optimization pathology in visual latent space contributions to predictions.

Ax Frederic Grabowski, Jacek Szczerbi\'nski, Maciej Ja\'skowski, Kalina Jasi\'nska-Kobus, Pawe{\l} D\k{a}browski-Tuma\'nski, Tomasz Jetka, Bartosz Topolski 5/5/2026

Bolek: A Multimodal Language Model for Molecular Reasoning

Multimodal language model for molecular property prediction that grounds explanations in molecular structure using Morgan fingerprints.

Ax Rufeng Chen, Zhaofan Zhang, Zhejiang Yang, Hechang Chen, Sihong Xie 5/5/2026

A decoupled diffusion planner that adapts to changing cost limits by using cost-conditioned generation for safety and reward gradients for performance

Offline safe reinforcement learning method using diffusion-based planners with cost-conditioned generation for adaptive safety constraints during deployment.

Ax Arian Eamaz, Farhang Yeganegi, Mojtaba Soltanalian 5/5/2026

Trust, but Verify: Peeling Low-Bit Transformer Networks for Training Monitoring

Layer-wise monitoring approach for transformer training using low-bit quantization to verify optimization quality without performance degradation.

Ax Lingxiao Kong, Cong Yang, Oya Deniz Beyan, Zeyd Boukhers 5/5/2026

Enhancing RL Generalizability in Robotics through SHAP Analysis of Algorithms and Hyperparameters

SHAP-based analysis decomposing RL algorithm and hyperparameter contributions to generalization across robotic environments.

Ax Shikhar Shukla 5/5/2026

SpecKV: Adaptive Speculative Decoding with Compression-Aware Gamma Selection

SpecKV: adaptive speculative decoding for LLM inference with compression-aware dynamic gamma selection outperforming fixed speculation.

Ax Dudekula Kasim Vali 5/5/2026

Autonomous QA Agent: A Retrieval-Augmented Framework for Reliable Selenium Script Generation

RAG-based autonomous QA agent using LLMs with retrieval grounding to generate reliable Selenium test scripts from documentation.

Ax Xiujiang Tan (Guangzhou Academy of Fine Arts, Guangzhou, China) 5/5/2026

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

Philosophical analysis of topological limitations in multimodal AI architectures regarding modal separability and creative tasks.

Ax Krisanu Sarkar 5/5/2026

Synthetic Designed Experiments for Diagnosing Vision Model Failure

Synthetic data generation approach diagnosing vision model failures through controlled, independent scene factor variation.

Ax Jay Lee, Hanqi Su, Marco Macchi, Adalberto Polenghi, Wei Wu, Zhiheng Zhao, George Q. Huang, Kiva Allgood, Devendra Jain, Benedikt Gieger, Vibhor Pandhare, Soumyabrata Bhattacharjee, Ram Mohril, Lingbao Kong, Qiyuan Wang, Xinlan Tang, Sungjong Kim, Chan Hee Park, Byeng D. Youn, Guo Dong Goh, Xi Huang, Wai Yee Yeong, Yung C Shin, He Zhang, Zitong Wang, Fei Tao, Jagjit Singh Srai, Satyandra K. Gupta, Byung Gun Joung, Albin John, John W. Sutherland, Sang Won Lee, Olga Fink, Vinay Sharma, Faez Ahmed, Wei Chen, Mark Fuge, Arild Waaler, Martin G. Skj{\ae}veland, Dimitris Kyritsis, Wei Chen, VispiNevile Karkaria, Yi-Ping Chen, Ying-Kuan Tsai, Joseph Cohen, Xun Huan, Jing Lin, Liangwei Zhang, Gregory W. Vogl, Aaron W. Cornelius, Xiaodong Jia, Dai-Yan Ji, Takanobu Minami, Ruoxin Wang 5/5/2026

2026 Roadmap on Artificial Intelligence and Machine Learning for Smart Manufacturing

2026 roadmap reviewing AI/ML deployment in smart manufacturing, covering challenges in data integration and industrial systems.

Ax Gouki Minegishi, Hiroki Furuta, Takeshi Kojima, Yusuke Iwasawa, Yutaka Matsuo 5/5/2026

Understanding Emergent Misalignment via Feature Superposition Geometry

Analysis of emergent misalignment in fine-tuned LLMs via feature superposition geometry, explaining harmful behavior induction.

Ax Cutter Dawes, Aryan Sharma, Angelos Ioannis Lagos, Shivam Raval 5/5/2026

H-Probes: Extracting Hierarchical Structures From Latent Representations of Language Models

H-Probes: linear probes extracting hierarchical structures from LLM latent representations to understand geometric reasoning.

Ax Firat Ozdemir, Yun Cheng, Salman Mohebi, Fanny Lehmann, Simon Adamov, Zhenyi Zhang, Leonardo Trentini, Dana Grund, Oliver Fuhrer, Torsten Hoefler, Siddhartha Mishra, Sebastian Schemm, Benedikt Soja, Mathieu Salzmann 5/5/2026

Earth System Foundation Model (ESFM): A unified framework for heterogeneous data integration and forecasting

Open Earth System Foundation Model (ESFM) built on Swin UNet for heterogeneous climate/weather data integration and forecasting.

Ax Peiliang Gong, Han Zhang, Zhen Jiang, Chenyu Liu, Ziyu Jia, Xinliang Zhou, Daoqiang Zhang, Xiaoli Li 5/5/2026

Foundation Model Guided Dual-Branch Co-Adaptation for Source-Free EEG Decoding

Foundation model-guided domain adaptation for EEG decoding without source data access, improving cross-subject generalization.

Ax Alizishaan Khatri, Chiquita Prabhu 5/5/2026

Latent Space Probing for Adult Content Detection in Video Generative Models

Latent space probing framework for detecting adult content in video generative models by analyzing internal representations.

Ax Yida Xue, Ningyu Zhang, Tingwei Wu, Zhe Ma, Daxiong Ji, Zhao Wang, Guozhou Zheng, Huajun Chen 5/5/2026

OceanPile: A Large-Scale Multimodal Ocean Corpus for Foundation Models

OceanPile: large-scale multimodal dataset for ocean research addressing data fragmentation and weak labeling challenges for foundation models.

Ax Ikhlas Enaieh (S2A, LTCI), Olivier Fercoq (S2A, LTCI), Garc\'ia \'Angel (DATSI, UPM) 5/5/2026

On the explainability of max-plus neural networks

Investigation of explainability in linear-min-max neural networks which can be interpreted as k-medoids and trained with subgradient descent.

Ax James Flora, Kowshik Thopalli, Akshay R. Kulkarni, Weng-Keen Wong, Shusen Liu 5/5/2026

LatentDiff: Scaling Semantic Dataset Comparison to Millions of Images

LatentDiff: scalable framework for semantic dataset comparison using sparse autoencoders and density ratio estimation in latent space.

Ax Hongjun Wang, Po Hu, Kai Han 5/5/2026

Generalized Category Discovery under Domain Shifts: From Vision to Vision-Language Models

Generalized Category Discovery frameworks adapting foundation models for unlabeled data with domain and semantic shifts.

Ax Han Gong, Zhen Zhou, Yunyang Shi, Yan Tan, Jinbiao Huo, Qi Hong, Zhiyuan Liu 5/5/2026

TRIP-Evaluate: An Open Multimodal Benchmark for Evaluating Large Models in Transportation

TRIP-Evaluate: open multimodal benchmark for assessing LLMs and MLLMs on transportation tasks including regulation QA and autonomous driving.

Ax Gourisetty Venkata Sai Koushik, Dama Aditya, Mahankali Harish Sai, Peddi Siddarhta, Shadab Ahmad, Vivek Yelleti 5/5/2026

PPO guided Agentic Pipeline for Adaptive Prompt Selection and Test Case Generation

Reinforcement learning agentic framework using PPO and LLM for automated test case generation in complex software systems.

Ax Kevin H. Guo, Chao Yan, Avinash Baidya, Katherine Brown, Xiang Goa, Juming Xiong, Zhijun Yin, Bradley A. Malin 5/5/2026

CLEAR: Revealing How Noise and Ambiguity Degrade Reliability in LLMs for Medicine

CLEAR framework evaluating how noise and ambiguity in decision-space presentation affect LLM reliability on medical reasoning tasks.

Ax Iman Sharifi, Hyeong Tae Kim, Maheed Hatem Ahmed, Mahsa Ghasemi, Peng Wei 5/5/2026

Separation Assurance between Heterogeneous Fleets of Small Unmanned Aerial Systems via Multi-Agent Reinforcement Learning

Multi-agent reinforcement learning for tactical deconfliction among heterogeneous fleets of unmanned aerial systems in dense airspace.

Ax Joseph Spracklen, Pedram Aghazadeh, Farinaz Koushanfar, Murtuza Jadliwala 5/5/2026

LLM Ghostbusters: Surgical Hallucination Suppression via Adaptive Unlearning

LLM Ghostbusters: Adaptive unlearning technique to suppress hallucinations in code generation, addressing supply-chain vulnerabilities from fictional packages.

Ax Zihao Yang, Mosh Levy, Yoav Goldberg, Byron C. Wallace 5/5/2026

Compared to What? Baselines and Metrics for Counterfactual Prompting

Framework for evaluating counterfactual prompting in LLMs by accounting for meaning-preserving baseline modifications to isolate causal factors.

Ax Shashank Kapadia, Deep Narayan Mishra, Sujal Reddy Alugubelli, Ajay Kumar, Swapnil Yadav, Rishi Bhatia 5/5/2026

SURGE: SuperBatch Unified Resource-efficient GPU Encoding for Heterogeneous Partitioned Data

SURGE: Production GPU encoding system for generating embeddings across 800M texts in 40K partitions with optimized inter-process communication.

Ax Lingxi Zhang, Guangtao Zheng, Hanjie Chen 5/5/2026

When Embedding-Based Defenses Fail: Rethinking Safety in LLM-Based Multi-Agent Systems

Analysis of safety vulnerabilities in LLM-based multi-agent systems where malicious agents exploit communication to propagate misinformation.

Ax Pratyush Muthukumar, Harshil Kotamreddy, Sarah Amiraslani, Tomo Kanazawa, Ramani Akkati, Shaan Jain, Andrew Mathau 5/5/2026

Multimodal Data Curation Through Ranked Retrieval

Data curation framework using ranked retrieval to improve multimodal embedding spaces by addressing modality bias and noisy paired supervision.

Ax Ghadi Nehme, Eamon Whalen, Faez Ahmed 5/5/2026

CADFit: Precise Mesh-to-CAD Program Generation with Hybrid Optimization

CADFit: Hybrid optimization method to generate parametric CAD programs from meshes using neural networks and design constraints.

Ax Kushal Jasti, Tejamani Prashanth Sahu, Rishitha Pentyala, Muvvala Mohit, Vivek Yelleti 5/5/2026

FeedbackLLM: Metadata driven Multi-Agentic Language Agnostic Test Case Generator with Evolving prompt and Coverage Feedback

FeedbackLLM: Multi-agent LLM system for automated test case generation using evolving prompts and coverage feedback to reduce hallucinations.

Ax Muyang Li, Yucheng Liu, Jianbo Ma, Elliot Osborne, Bo Han, Tongliang Liu 5/5/2026

Rethinking Model Selection in VLM Through the Lens of Gromov-Wasserstein Distance

Systematic investigation of vision encoder-LLM alignment in Vision-Language Models using Gromov-Wasserstein distance for principled model selection.

Ax Lei Gao, Zhuoming Li, Mengxi Jia, Jiakang Yuan, Hongbo Sun, Hao Sun, Xuelong Li 5/5/2026

Segment-Aligned Policy Optimization for Multi-Modal Reasoning

Segment-Aligned Policy Optimization (SAPO) for LLM reinforcement learning that aligns credit assignment with reasoning step structure in multi-modal tasks.

Ax Anjie Liu, Ziqin Gong, Yan Song, Yuxiang Chen, Xiaolong Liu, Hengtong Lu, Kaike Zhang, Chen Wei 5/5/2026

Active Reasoning Vision-Language Models via Sequential Experimental Design

Vision-language models with active reasoning via sequential Bayesian decision-making. VLM improvement through adaptive visual perception.

Ax Jianze Wang, Ying Liu, Jinlong Chen, Xuchun Hu, Qilong Zhang, Yu Cao, Jun Wang, Hua Yang, Yong Xie, Qianglong Chen 5/5/2026

MAD-OPD: Breaking the Ceiling in On-Policy Distillation via Multi-Agent Debate

Multi-agent debate for on-policy distillation in agentic tasks. Teacher-student learning framework with agent trajectory optimization.

Ax Burin Naowarat, Hao Tang, Sharon Goldwater 5/5/2026

A framework for analyzing concept representations in neural models

Framework for analyzing concept representations in neural networks via linear subspaces. Interpretability research applicable to model understanding.

Ax Yao Du, Shanshan Li, Xiaomeng Li 5/5/2026

Injecting Distributional Awareness into MLLMs via Reinforcement Learning for Deep Imbalanced Regression

Using RL to improve MLLMs on imbalanced regression tasks via distributional awareness. LLM training methodology addressing long-tailed distributions.

Ax Shuaipeng Zhou, Yu Zhang 5/5/2026

SCALE-LoRA: Auditing Post-Retrieval LoRA Composition with Residual Merging and View Reliability

Framework for composing and auditing LoRA adapters from open pools for tasks. Parameter-efficient fine-tuning and model composition.

Ax Mehmet Iscan 5/5/2026

Feedback-Normalized Developer Memory for Reinforcement-Learning Coding Agents: A Safety-Gated MCP Architecture

LLM coding agents with persistent memory, RAG, and RL feedback for software engineering. Architecture for agent memory and tool use.

Ax Tingting Dan, Guorong Wu 5/5/2026

From Cortical Synchronous Rhythm to Brain Inspired Learning Mechanism: An Oscillatory Spiking Neural Network with Time-Delayed Coordination

Brain-inspired spiking neural network with time-delayed coordination for learning. Neuroscience-focused theoretical work on oscillatory dynamics.

Ax Xihang Shan, Da Zhou 5/5/2026

PRCD-MAP: Learning How Much to Trust Imperfect Priors in Causal Discovery

Causal discovery method with per-edge trust scores for heterogeneously reliable external priors from diverse sources.