Isolater - Feed

Ax Sebastian Wind, Tri-Thien Nguyen, Jeta Sopa, Mahshad Lotfinia, Sebastian Bickelhaup, Michael Uder, Harald K\"ostler, Gerhard Wellein, Sven Nebelung, Daniel Truhn, Andreas Maier, Soroosh Tayebi Arasteh 5/6/2026

Safety and accuracy follow different scaling laws in clinical large language models

Framework measuring how safety and accuracy scale differently in clinical LLMs, showing size increases don't guarantee safer behavior.

Ax Minh H. Vu, Daniel Edler, Carl Wibom, Tommy L\"ofstedt, Beatrice Melin, Martin Rosvall 5/6/2026

A Unified Framework for Tabular Generative Modeling: Loss Functions, Benchmarks, and Improved Multi-objective Bayesian Optimization Approaches

Unified framework for tabular generative modeling with loss functions, benchmarks, and multi-objective Bayesian optimization approaches.

Ax Yifei Li, Erik-Jan van Kampen 5/6/2026

Deep deterministic policy gradient with symmetric data augmentation for lateral attitude tracking control of a fixed-wing aircraft

Applies symmetric data augmentation with deep deterministic policy gradient for aircraft control. Uses reinforcement learning with sample efficiency.

Ax Itamar Zimerman, Allon Adir, Ehud Aharoni, Matan Avitan, Moran Baruch, Nir Drucker, Jenny Lerner, Ramy Masalha, Reut Meiri, Omri Soceanu 5/6/2026

Power-Softmax: Towards Secure LLM Inference over Encrypted Data

Method for privacy-preserving LLM inference over encrypted data by approximating Softmax and layer normalization as polynomials for homomorphic encryption.

Ax Miaomiao Li, Hao Chen, Yang Wang, Tingyuan Zhu, Weijia Zhang, Kaijie Zhu, Kam-Fai Wong, Jindong Wang 5/6/2026

Understanding and Mitigating Bias Inheritance in LLM-based Data Augmentation on Downstream Tasks

Studies bias inheritance when LLMs generate synthetic training data, showing how biases propagate and amplify in downstream tasks.

Ax Pranjal Awasthi, Sreenivas Gollapudi, Ravi Kumar, Kamesh Munagala 5/6/2026

Sample-Efficient Optimization over Generative Priors via Coarse Learnability

Research on zeroth-order optimization with generative priors using coarse learnability for sample-efficient model-based optimization.

Ax Nathanael Tepakbong, Xiang Zhou, Ding-Xuan Zhou 5/6/2026

Super-fast Rates of Convergence for Neural Network Classifiers under the Hard Margin Condition

Convergence rate analysis for deep neural network classifiers under hard margin condition.

Ax Mohamed Amine Ketata, David L\"udke, Leo Schwinn, Stephan G\"unnemann 5/6/2026

Joint Relational Database Generation via Graph-Conditional Diffusion Models

Graph-conditional diffusion models for joint generation of multi-table relational databases.

Ax Noah Amsel, David Persson, Christopher Musco, Robert M. Gower 5/6/2026

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

Optimal matrix sign computation methods for neural network training via the Muon optimizer.

Ax Jiaru Zhang, Juanwu Lu, Xiaoyu Wu, Ziran Wang, Ruqi Zhang 5/6/2026

Accelerating Inference of Discrete Autoregressive Normalizing Flows by Selective Jacobi Decoding

Selective Jacobi decoding accelerates discrete autoregressive normalizing flows for generative modeling.

Ax Mehul Bafna, Siddhant anand Jadhav, David Sweet 5/6/2026

Taking the GP Out of the Loop

Scaling Bayesian optimization beyond Gaussian processes for problems with abundant cheap observations.

Ax Minjae Lee, Yoonjae Jung, Sangdon Park 5/6/2026

Online Conformal Abstention for Factuality Control Under Adversarial Bandit Feedback

Online conformal abstention for LLM factuality control with adversarial bandit feedback in interactive systems.

Ax Vishnu Vinod, Krishna Pillutla, Abhradeep Guha Thakurta 5/6/2026

InvisibleInk: High-Utility and Low-Cost Text Generation with Differential Privacy

InvisibleInk framework for differentially private long-form text generation with RAG and inference-time scaling.

Ax Joydeep Chandra, Prabal Manhas, Ramanjot Kaur, Rashi Sahay 5/6/2026

Aura-CAPTCHA: A Reinforcement Learning and GAN-Enhanced Multi-Modal CAPTCHA System

CAPTCHA system combining GANs and RL for adaptive multi-modal verification challenges.

Ax Qiyu Chen, Guozhang Chen 5/6/2026

Aligning Inductive Bias for Data-Efficient Generalization in State Space Models

Aligning inductive bias in state space models for improved data-efficient generalization.

Ax Larissa Reichart, Cem Ata Baykara, Ali Burak \"Unal, Harlin Lee, Mete Akg\"un 5/6/2026

Scaling Unsupervised Multi-Source Federated Domain Adaptation through Group-Wise Discrepancy Minimization

Federated unsupervised multi-source domain adaptation framework addressing scalability and privacy.

Ax Zhihan Zhang, Xunkai Li, Yilong Zuo, Henan Sun, Zhenjun Li, Bing Zhou, Rong-Hua Li, Guoren Wang 5/6/2026

When LLM Agents Meet Graph Optimization: An Automated Data Quality Improvement Approach

LLM agents automated data quality improvement for text-attributed graphs using graph optimization techniques.

Ax Khang Ngo, Siamak Ravanbakhsh 5/6/2026

Scaling Laws and Symmetry, Evidence from Neural Force Fields

Empirical study showing equivariant neural architectures scale better than non-equivariant models for interatomic potential learning.

Ax Aviv Ratzon, Omri Barak 5/6/2026

Prediction horizon shapes representations in predictive learning

Study of how prediction horizon affects learned representations in predictive models and world models.

Ax Prabodh Katti, Houssem Sifaou, Sangwoo Park, Bipin Rajendran, Osvaldo Simeone 5/6/2026

On-Device Fine-Tuning via Backprop-Free Zeroth-Order Optimization

Zeroth-order optimization for on-device fine-tuning of AI agents without backpropagation, reducing memory constraints.

Ax Nathan M. Roberts II, Xiaosong Du 5/6/2026

Transformer-Guided Deep Reinforcement Learning for Optimal Takeoff Trajectory Design of an eVTOL Drone

Deep RL with Transformers optimizes eVTOL aircraft takeoff trajectories for minimum energy consumption.

Ax Dongseok Kim, Hyoungsun Choi, Mohamed Jismy Aashik Rasool, Gisung Oh 5/6/2026

CLAPS: Aleatoric-Epistemic Scaling via Last-Layer Laplace for Conformal Regression

Conformal regression method using Laplace approximation for adaptive prediction intervals with uncertainty quantification.

Ax Prakhar Gupta, Vaibhav Gupta 5/6/2026

Bootstrapped Mixed Rewards for RL Post-Training: Injecting Canonical Action Order

RL post-training with canonical action order hints improves Transformer performance on constraint satisfaction puzzles like Zebra puzzles.

Ax Chen Liu, Xingzhi Sun, Xi Xiao, Alexandre Van Tassel, Ke Xu, Kristof Reimann, Danqi Liao, Mark Gerstein, Tianyang Wang, Xiao Wang, Smita Krishnaswamy 5/6/2026

Dispersion Loss Counteracts Embedding Condensation and Improves Generalization in Small Language Models

Analysis of embedding condensation in small LLMs and dispersion loss technique to improve generalization.

Ax Jinbin Bai, Yixuan Li, Yuchen Zhu, Yi Xin, Qingyu Shi, Aosong Feng, Xiaohong Liu, Molei Tao, Jianru Xue, Xiangtai Li, Ming-Hsuan Yang 5/6/2026

Prism: Efficient Test-Time Scaling via Hierarchical Search and Self-Verification for Discrete Diffusion Language Models

Prism: test-time scaling method for discrete diffusion language models using hierarchical search and self-verification.

Ax Junchen Liu, Sven Elflein, Or Litany, Zan Gojcic, Ruilong Li 5/6/2026

Test-Time Training with KV Binding Is Secretly Linear Attention

Analysis showing test-time training with KV binding operates as learned linear attention, not memorization.

Ax Yu Zhang, Songwei Liu, Chenqian Yan, Sheng Lin, Beichen Ning, Fangmin Chen, Xing Wang 5/6/2026

S2O: Early Stopping for Sparse Attention via Online Permutation

S2O algorithm enabling fine-grained sparse attention via online permutation for efficient long-context inference.

Ax Tanishq Kumar, Tri Dao, Avner May 5/6/2026

Speculative Speculative Decoding

Speculative speculative decoding technique accelerating LLM inference by parallelizing draft model speculation.

Ax Zonglin Yang, Lidong Bing 5/6/2026

MOOSE-Star: Unlocking Tractable Training for Scientific Discovery by Breaking the Complexity Barrier

Method for tractable training of LLMs for scientific discovery by decomposing hypothesis generation complexity.

Ax Hyunji Nam, Haoran Li, Natasha Jaques 5/6/2026

Maximizing mutual information between prompts and responses improve LLM personalization with no additional data or human oversight

Self-improvement framework for LLM personalization via mutual information maximization without additional labeled data.

Ax Yubai Wei, Chen Wu, Hashem Haghbayan 5/6/2026

Can Explicit Physical Feasibility Benefit VLA Learning? An Empirical Study

Empirical study on incorporating explicit physical constraints into Vision-Language-Action robot models.

Ax Donghwan Lee 5/6/2026

Lyapunov-Certified Direct Switching Theory for Q-Learning

Theoretical framework analyzing Q-learning through switching systems and Bellman error representation.

Ax Kang Liu, Jianchen Hu, Wei Peng 5/6/2026

SOC-ICNN: From Polyhedral to Conic Geometry for Learning Convex Surrogate Functions

Extends input convex neural networks from linear programming to second-order cone programming for improved representational capacity of surrogate functions.

Ax Isaac Tosin Adisa 5/6/2026

An Integrated Framework for Explainable, Fair, and Observable Hospital Readmission Prediction: Development and Validation on MIMIC-IV

Hospital readmission prediction framework addressing explainability, deployment reliability, and fairness using MIMIC-IV clinical database.

Ax A. Yermekov, D. A. Herrera-Mart\'i 5/6/2026

StackFeat RL: Reinforcement Learning over Iterative Dual Criterion Feature Selection for Stable Biomarker Discovery

Reinforcement learning meta-framework for stable biomarker discovery through iterative feature selection in high-dimensional genomic data.

Ax Gabe Guo, Thanawat Sornwanee, Lutong Hao, Elon Litman, Stefano Ermon, Jose Blanchet 5/6/2026

ABC: Any-Subset Autoregression via Non-Markovian Diffusion Bridges in Continuous Time and Space

Diffusion-based approach for generating continuous-time stochastic processes conditioned on partial observations using non-Markovian bridges.

Ax Daniel Zhu, Zihan Wang, Xuchan Bao, Jerry Wei 5/6/2026

Jailbroken Frontier Models Retain Their Capabilities

Analysis showing advanced jailbreaks of frontier language models impose minimal performance degradation compared to less complex attacks.

Ax Zhi-Quan Feng, Ying-Jia Lin, Hung-Yu Kao 5/6/2026

Learning in the Fisher Subspace: A Guided Initialization for LoRA Fine-Tuning

LoRA fine-tuning method using Fisher subspace initialization to improve LLM adaptation efficiency and downstream task performance.

Ax Youssef Abo-Dahab, Xiaoiang Xiang, Joanne Chun, Liang Zhao 5/6/2026

Benchmarking Single-Pose Docking, Consensus Rescoring, and Supervised ML on the LIT-PCBA Library: A Critical Evaluation of DiffDock, AutoDock-GPU, GNINA, and DiffDock-NMDN

Large-scale benchmark evaluating docking and scoring methods including AI-based DiffDock and NMDN on LIT-PCBA library for virtual screening.

Ax Fernando Martin-Maroto, Nabil Abderrahaman, Gonzalo G. de Polavieja 5/6/2026