Ax Gonzalo N\'apoles, Isel Grau, Yamisleydi Salgueiro 4/22/2026

Concept Inconsistency in Dermoscopic Concept Bottleneck Models: A Rough-Set Analysis of the Derm7pt Dataset

Analyzes concept inconsistencies in dermoscopic concept bottleneck models using rough set theory on medical imaging dataset.

Ax Simin Yu, Sufia Fathima 4/22/2026

When Active Learning Falls Short: An Empirical Study on Chemical Reaction Extraction

Empirical study examining active learning for chemical reaction extraction from literature, identifying annotation cost barriers.

Ax Harekrushna Sahu, Pratik Jawanpuria, Pranay Sharma 4/22/2026

FedSEA: Achieving Benefit of Parallelization in Federated Online Learning

Extends online federated learning with stochastic adversary model capturing statistical variation and enabling parallelization benefits.

Ax Haotian Ye, Haowei Lin, Jingyi Tang, Yizhen Luo, Caiyin Yang, Chang Su, Rahul Thapa, Rui Yang, Ruihua Liu, Zeyu Li, Chong Gao, Dachao Ding, Guangrong He, Miaolei Zhang, Lina Sun, Wenyang Wang, Yuchen Zhong, Zhuohao Shen, Di He, Jianzhu Ma, Stefano Ermon, Tongyang Li, Xiaowen Chu, James Zou, Yuzhi Xu 4/22/2026

Evaluation-driven Scaling for Scientific Discovery

Formulates language model scaling for scientific discovery around evaluation-driven loops with verifiers and task-specific scoring functions.

Ax Huayu Deng, Jinghui Zhong, Xiangming Zhu, Yunbo Wang, Xiaokang Yang 4/22/2026

LASER: Learning Active Sensing for Continuum Field Reconstruction

Proposes LASER framework formulating active sensing for field reconstruction as POMDP with closed-loop adaptive sensor placement.

Ax Rudolf Debelak 4/22/2026

FairTree: Subgroup Fairness Auditing of Machine Learning Models with Bias-Variance Decomposition

Introduces FairTree algorithm for auditing machine learning models for subgroup fairness using bias-variance decomposition framework.

Ax Vasiliki Papanikou, Evaggelia Pitoura 4/22/2026

TACENR: Task-Agnostic Contrastive Explanations for Node Representations

Proposes TACENR for task-agnostic explainability of node representations in graph neural networks using contrastive methods.

Ax Yi Zhao, Di Yuan, Tao Deng, Suzhi Cao, Ying Dong 4/22/2026

Optimal Routing for Federated Learning over Dynamic Satellite Networks: Tractable or Not?

Studies optimal routing for federated learning over dynamic satellite networks with relay-based communication between clients and servers.

Ax Gerard Pons, Carlos Escolano, Besim Bilalli, Anna Queralt 4/22/2026

Revisiting Catastrophic Forgetting in Continual Knowledge Graph Embedding

Analyzes catastrophic forgetting in continual knowledge graph embedding updates and proposes approaches beyond limiting embedding changes.

Ax Thomas Zollo, Jimmy Wang, Richard Zemel 4/22/2026

Unsupervised Confidence Calibration for Reasoning LLMs from a Single Generation

Introduces unsupervised method for calibrating confidence estimates in reasoning LLMs using single generation without labels or repeated sampling.

Ax Yuhan Hu, Xiaolei Fang 4/22/2026

Heterogeneity-Aware Personalized Federated Learning for Industrial Predictive Analytics

Proposes personalized federated learning approach for industrial predictive analytics accounting for heterogeneous degradation processes across clients.

Ax Chengjun Pan, Shichun Liu, Jiahang Lin, Dingwei Zhu, Jiazheng Zhang, Shihan Dou, Songyang Gao, Zhenhua Han, Binghai Wang, Rui Zheng, Xuanjing Huang, Tao Gui, Yansong Feng 4/22/2026

EVPO: Explained Variance Policy Optimization for Adaptive Critic Utilization in LLM Post-Training

Proposes EVPO method for LLM post-training via reinforcement learning that optimizes critic utilization in policy gradient optimization.

Ax Saket Maganti 4/22/2026

When Graph Structure Becomes a Liability: A Critical Re-Evaluation of Graph Neural Networks for Bitcoin Fraud Detection under Temporal Distribution Shift

Evaluates GNN performance on Bitcoin fraud detection under strict inductive protocols, finding random forests outperform GCN, GraphSAGE, and GAT methods.

Ax Ziqin Chen, Zuang Wang, Yongqiang Wang 4/22/2026

Accelerating Optimization and Machine Learning through Decentralization

Reviews decentralized optimization approaches for distributed machine learning across devices with local datasets, emphasizing privacy and scalability benefits.

Ax Jianyang Gao, Yutong Gou, Yuexuan Xu, Jifan Shi, Yongyi Yang, Shuolin Li, Raymond Chi-Wing Wong, Cheng Long 4/22/2026

Revisiting RaBitQ and TurboQuant: A Symmetric Comparison of Methods, Theory, and Experiments

Comparative analysis of RaBitQ and TurboQuant quantization methods with reproducible experimental framework showing inconsistent performance differences.

Ax Akash Yadav, Taiwo A. Adebiyi, Ruda Zhang 4/22/2026

Calibrating Scientific Foundation Models with Inference-Time Stochastic Attention

Proposes stochastic attention modification for transformer-based scientific models to improve calibrated uncertainty estimates at inference time.

Ax Maxim Raginsky, Benjamin Recht 4/22/2026

Separating Geometry from Probability in the Analysis of Generalization

Theoretical analysis separating geometry from probability in machine learning generalization bounds for in-sample and out-of-sample performance.

Ax Carles Navarro, Philipp Tholke, Gianni de Fabritiis 4/22/2026

Structure-guided molecular design with contrastive 3D protein-ligand learning

Framework for structure-based drug discovery combining contrastive 3D protein-ligand learning with autoregressive molecular generation on commercial compounds.

Ax Donghwan Lee 4/22/2026

Lyapunov-Certified Direct Switching Theory for Q-Learning

Theoretical analysis of constant-stepsize Q-learning using stochastic switching systems and Lyapunov methods for convergence certification.

Ax Gabriele Farina, Juan Carlos Perdomo 4/22/2026

An Efficient Black-Box Reduction from Online Learning to Multicalibration, and a New Route to $\Phi$-Regret Minimization

Black-box reduction from online learning to multicalibration using no-regret learners and expected variational inequality solvers.

Ax Inhyeok Choi, Hyuncheol Park 4/22/2026

SAGE: Training-Free Semantic Evidence Composition for Edge-Cloud Inference under Hard Uplink Budgets

SAGE method for edge-cloud hybrid inference under hard uplink bandwidth constraints using semantic evidence composition instead of attention-based importance.

Ax Xudong Jian, Charikleia Stoura, Simon Scandella, Eleni Chatzi 4/22/2026

Disentangling Damage from Operational Variability: A Label-Free Self-Supervised Representation Learning Framework for Output-Only Structural Damage Identification

Self-supervised representation learning framework for structural damage identification that disentangles damage from operational variability without labels.

Ax Andrea Goertzen, Kaveh Alim, Navid Azizan 4/22/2026

HardNet++: Nonlinear Constraint Enforcement in Neural Networks

HardNet++ enforces nonlinear constraints in neural network outputs for control and decision-making with guaranteed constraint satisfaction at inference.

Ax Salvatore Greco, Jacek Karolczak, Roman S{\l}owi\'nski, Jerzy Stefanowski 4/22/2026

PREF-XAI: Preference-Based Personalized Rule Explanations of Black-Box Machine Learning Models

PREF-XAI framework generates personalized rule-based explanations for black-box ML models based on user preferences and cognitive constraints.

Ax Jake Lee 4/22/2026

Adaptive MSD-Splitting: Enhancing C4.5 and Random Forests for Skewed Continuous Attributes

Adaptive MSD-Splitting algorithm improves discretization of continuous attributes in decision trees like C4.5 and Random Forests for efficiency and accuracy.

Ax Jiaming Zhang, Meng Ding, Shaopeng Fu, Jingfeng Zhang, Di Wang 4/22/2026

Benign Overfitting in Adversarial Training for Vision Transformers

Theoretical analysis of adversarial training in Vision Transformers, studying robustness to adversarial examples and benign overfitting.

Ax Abdulmoneam Ali, Ahmed Arafa 4/22/2026

FB-NLL: A Feature-Based Approach to Tackle Noisy Labels in Personalized Federated Learning

FB-NLL: Feature-based approach handling noisy labels in personalized federated learning with heterogeneous data.

Ax Perry Dong, Alexander Swerdlow, Dorsa Sadigh, Chelsea Finn 4/22/2026

FASTER: Value-Guided Sampling for Fast RL

FASTER: Value-guided sampling method reducing computational cost of test-time scaling in diffusion-based RL policies.

Ax Austin Coursey, Abel Diaz-Gonzalez, Marcos Quinones-Grueiro, Gautam Biswas 4/22/2026

Safe Continual Reinforcement Learning in Non-stationary Environments

Safe continual RL for non-stationary environments ensuring constraint satisfaction during learning in physical control systems.

Ax Joongwon Kim, Wannan Yang, Kelvin Niu, Hongming Zhang, Yun Zhu, Eryk Helenowski, Ruan Silva, Zhengxing Chen, Srinivasan Iyer, Manzil Zaheer, Daniel Fried, Hannaneh Hajishirzi, Sanjeev Arora, Gabriel Synnaeve, Ruslan Salakhutdinov, Anirudh Goyal 4/22/2026

Scaling Test-Time Compute for Agentic Coding

Method for scaling test-time compute in agentic coding systems using trajectory ranking and value estimates for long-horizon tasks.

Ax Logan Halle, Jason P. Gleghorn 4/22/2026

Dual Triangle Attention: Effective Bidirectional Attention Without Positional Embeddings

Dual Triangle Attention: Bidirectional transformer attention mechanism eliminating need for positional embeddings.

Ax Xudong Wang, Chaoning Zhang, Chenghao Li, Shuxu Chen, Qigan Sun, Jiaquan Zhang, Fachrina Dewi Puspitasari, Tae-Ho Kim, Jiwei Wei, Malu Zhang, Guoqing Wang, Yang Yang, Heng Tao Shen 4/22/2026

Agent-GWO: Collaborative Agents for Dynamic Prompt Optimization in Large Language Models

Agent-GWO: Multi-agent collaborative system for dynamic prompt optimization in LLMs using genetic wolf optimizer.

Ax Ruixuan Liu, David Evans, Li Xiong 4/22/2026

Beyond Indistinguishability: Measuring Extraction Risk in LLM APIs

Analysis showing indistinguishability properties don't prevent data extraction from LLM APIs; formalizes privacy game separation.

Ax Hanrui Luo, Shreyank N Gowda 4/22/2026

An Empirical Study of Multi-Generation Sampling for Jailbreak Detection in Large Language Models

Empirical study of jailbreak detection in LLMs using multi-generation sampling on JailbreakBench with varying model alignment.

Ax Jiacheng Liang, Yao Ma, Tharindu Kumarage, Satyapriya Krishna, Rahul Gupta, Kai-Wei Chang, Aram Galstyan, Charith Peris 4/22/2026

ARES: Adaptive Red-Teaming and End-to-End Repair of Policy-Reward System

ARES: Red-teaming and repair method for RLHF systems addressing vulnerabilities in both policy and reward model simultaneously.

Ax Marti\~no R\'ios-Garc\'ia, Nawaf Alampara, Chandan Gupta, Indrajeet Mandal, Sajid Mannan, Ali Asghar Aghajani, N. M. Anoop Krishnan, Kevin Maik Jablonka 4/22/2026

AI scientists produce results without reasoning scientifically

Evaluation of LLM-based scientific agents across 8 domains with 25k runs assessing whether they follow scientific reasoning norms.

Ax Sinan G. Aksoy, Alexandra A. Sabrio, Erik VonKaenel, Lee Burke 4/22/2026

Semantic Needles in Document Haystacks: Sensitivity Testing of LLM-as-a-Judge Similarity Scoring

Framework testing LLM sensitivity to semantic changes in document comparison tasks using perturbation-based needle-in-haystack experiments.

Ax Junhao Dong, Yifei Zhang, Hao Zhu, Yew-Soon Ong, Piotr Koniusz 4/22/2026

Hierarchically Robust Zero-shot Vision-language Models

Hierarchically robust zero-shot vision-language models defending against adversarial attacks on both base and superclass levels.

Ax Manuel Israel Cazares 4/22/2026

Less Is More: Cognitive Load and the Single-Prompt Ceiling in LLM Mathematical Reasoning

Empirical study of prompt engineering for formal mathematical reasoning, analyzing 40+ prompt variants in equational theory competition.

Ax Mehul Agarwal, Aditya Aggarwal, Arnav Goel, Medha Hira, Anubha Gupta 4/22/2026

MORPHOGEN: A Multilingual Benchmark for Evaluating Gender-Aware Morphological Generation

MORPHOGEN: multilingual benchmark for evaluating LLM handling of gender-aware morphological generation in 13 languages.

Ax Cristina Garbacea, Heran Wang, Chenhao Tan 4/22/2026

Personalized Benchmarking: Evaluating LLMs by Individual Preferences

Framework for personalized LLM benchmarking that evaluates alignment with individual user preferences instead of aggregate ratings.

Ax Zhenbang Du, Kejing Xia, Xinrui Zhong, Yonggan Fu, Nicolai Oswald, Binfei Ji, Brucek Khailany, Pavlo Molchanov, Yingyan Lin 4/22/2026

$R^2$-dLLM: Accelerating Diffusion Large Language Models via Spatio-Temporal Redundancy Reduction

R²-dLLM: acceleration technique for diffusion-based LLMs via spatio-temporal redundancy reduction during parallel token decoding.

Ax Yilun Liu, Ruihong Qiu, Zi Huang 4/22/2026

TRN-R1-Zero: Text-rich Network Reasoning via LLMs with Reinforcement Learning Only

TRN-R1-Zero: zero-shot reasoning on text-rich networks using LLMs with reinforcement learning, integrating textual semantics with graph structure.

Ax Rania Elbadry, Sarfraz Ahmad, Ahmed Heakl, Dani Bouch, Momina Ahsan, Muhra AlMahri, Marwa Elsaid khalil, Yuxia Wang, Salem Lahlou, Sophia Ananiadou, Veselin Stoyanov, Jimin Huang, Xueqing Peng, Preslav Nakov, Zhuohan Xie 4/22/2026

SAHM: A Benchmark for Arabic Financial and Shari'ah-Compliant Reasoning

SAHM: benchmark and instruction-tuning dataset for Arabic financial NLP and Sharia-compliant reasoning with 14,380 examples.

Ax Zhenghua Ma, G Abarajithan, Dimitrios Danopoulos, Olivia Weng, Francesco Restuccia, Ryan Kastner 4/22/2026

Design Rules for Extreme-Edge Scientific Computing on AI Engines

Design principles for deploying machine learning models on extreme-edge AI hardware with latency/throughput constraints using spatial dataflow.

Ax Xiao Qi Lee, Ezinne Nwankwo, Angela Zhou 4/22/2026

Auditing LLMs for Algorithmic Fairness in Casenote-Augmented Tabular Prediction

Audits fairness properties of LLM-based tabular classification on housing placement prediction with real nonprofit casenote data, evaluating multi-class classification error disparities.

Ax Daniel M. Jimenez-Gutierrez, Enrique Zuazua, Georgios Kellaris, Joaquin Del Rio, Oleksii Sliusarenko, Xabi Uribe-Etxebarria 4/22/2026

Sherpa.ai Privacy-Preserving Multi-Party Entity Alignment without Intersection Disclosure for Noisy Identifiers

Privacy-preserving entity alignment method for vertical federated learning without disclosing intersection information using noisy identifiers.

Ax Abu Noman Md Sakib, Md. Main Oddin Chisty, Zijie Zhang 4/22/2026

Beyond Semantic Similarity: A Component-Wise Evaluation Framework for Medical Question Answering Systems with Health Equity Implications

Evaluation framework for medical question answering systems using LLMs, assessing accuracy and health equity implications beyond semantic similarity.

Ax Divyesh Gabbireddy, Suman Saha 4/22/2026

Evaluating LLM-Generated Obfuscated XSS Payloads for Machine Learning-Based Detection

Evaluation of LLM-generated obfuscated XSS payloads and their detection by machine learning-based security systems.

Ax Jonas Waldendorf, Bashar Awwad Shiekh Hasan, Evgenii Tsymbalov 4/22/2026

Detecting Hallucinations in SpeechLLMs at Inference Time Using Attention Maps

Method for detecting hallucinations in speech-based LLMs using attention map metrics without requiring gold-standard outputs.