Isolater - Feed

Ax Jacek Karwowski, Raymond Douglas 4/2/2026

Incoherence in Goal-Conditioned Autoregressive Models

Mathematical analysis of incoherence in goal-conditioned autoregressive models, studying policy improvement through fine-tuning with online RL.

Ax Giovanni Conforti, Alain Durmus, Le-Tuyet-Nhi Pham, Gael Raoul 4/2/2026

Non-Asymptotic Convergence of Discrete Diffusion Models: Masked and Random Walk dynamics

Theoretical analysis of diffusion models on discrete state spaces, establishing convergence guarantees for masked and random walk dynamics.

Ax Takuya Kanazawa 4/2/2026

Multivariate Uncertainty Quantification with Tomographic Quantile Forests

Tomographic Quantile Forests (TQF) for nonparametric uncertainty quantification in multivariate regression tasks.

Ax Kevin Zhang, Yixin Wang 4/2/2026

Meta-probabilistic Modeling

Meta-probabilistic modeling framework for discovering latent structure across collections of related datasets using probabilistic graphical models.

Ax Anderson de Andrade, Alon Harell, Ivan V. Baji\'c 4/2/2026

Lossy Common Information in a Learnable Gray-Wyner Network

Research on learnable Gray-Wyner networks for disentangling common and task-specific information in computer vision.

Ax Yuze Wang, Yujia Tong, Xuan Liu, Junhao Dong 4/2/2026

SAU: Sparsity-Aware Unlearning for LLMs via Gradient Masking and Importance Redistribution

SAU method for machine unlearning in sparse LLMs via gradient masking and importance redistribution for privacy.

Ax Sohan Venkatesh, Ashish Mahendran Kurapath 4/2/2026

On the Non-Identifiability of Steering Vectors in Large Language Models

Research showing activation steering vectors in LLMs are fundamentally non-identifiable with large equivalence classes.

Ax Isaac Han, Sangyeon Park, Seungwon Oh, Donghu Kim, Hojoon Lee, Kyung-Joong Kim 4/2/2026

FIRE: Frobenius-Isometry Reinitialization for Balancing the Stability-Plasticity Tradeoff

FIRE method for reinitialization in continual learning that balances stability and plasticity in neural networks.

Ax Deyi Kong, Zaiwei Chen, Shuzhong Zhang, Shancong Mou 4/2/2026

Natural Hypergradient Descent: Algorithm Design, Convergence Analysis, and Parallel Implementation

Research on Natural Hypergradient Descent for bilevel optimization using Fisher information matrix as Hessian surrogate.

Ax Tatsuya Sagawa, Ryosuke Kojima 4/2/2026

How Well Do Large-Scale Chemical Language Models Transfer to Downstream Tasks?

Evaluation of scaling laws for Chemical Language Models on downstream molecular property prediction tasks.

Ax YanZhao Zheng, ZhenTao Zhang, Chao Ma, YuanQiang Yu, JiHuai Zhu, Yong Wu, Tianze Xu, Baohua Dong, Hangcheng Zhu, Ruohui Huang, Gang Yu 4/2/2026

SkillRouter: Skill Routing for LLM Agents at Scale

Skill routing system for LLM agents that identifies relevant skills from large ecosystems before planning or execution.

Ax Dogan Urgun, Gokhan Gungor 4/2/2026

Large Language Model Guided Incentive Aware Reward Design for Cooperative Multi-Agent Reinforcement Learning

Framework using LLMs to automatically design reward programs for cooperative multi-agent RL systems with sparse task feedback.

Ax Pengxuan Yang, Yupeng Zheng, Deheng Qian, Zebin Xing, Qichao Zhang, Linbo Wang, Yichen Zhang, Shaoyu Guo, Zhongpu Xia, Qiang Chen, Junyu Han, Lingyun Xu, Yifeng Pan, Dongbin Zhao 4/2/2026

DreamerAD: Efficient Reinforcement Learning via Latent World Model for Autonomous Driving

DreamerAD uses latent world models for efficient RL in autonomous driving, compressing diffusion sampling 80x with visual interpretability.

Ax Marc-Antoine Allard, Arnaud Teinturier, Victor Xing, Gautier Viaud 4/2/2026

Experiential Reflective Learning for Self-Improving LLM Agents

ERL framework enabling LLM agents to self-improve through experiential learning from past interactions and reflective adaptation.

Ax Devashish Gaikwad, Wil M. P. van der Aalst, Gyunam Park 4/2/2026

Neuro-Symbolic Process Anomaly Detection

Neuro-symbolic method for process anomaly detection combining neural networks with domain knowledge from process mining.

Ax Yufei Xu, Fanxu Meng, Fan Jiang, Yuxuan Wang, Ruijie Zhou, Zhaohui Wang, Jiexi Wu, Zhixin Pan, Xiaojuan Tang, Wenjie Pei, Tongxuan Liu, Di yin, Xing Sun, Muhan Zhang 4/2/2026

HISA: Efficient Hierarchical Indexing for Fine-Grained Sparse Attention

Hierarchical indexing system for efficient fine-grained sparse attention in transformers, removing bottleneck from key selection.

Ax Max Hennick, Guillaume Corlouer 4/2/2026

From Density Matrices to Phase Transitions in Deep Learning: Spectral Early Warnings and Interpretability

Uses 2-datapoint reduced density matrix from quantum chemistry to predict and understand phase transitions during neural network training.

Ax Chinmay Savadikar, Michelle Dai, Tianfu Wu 4/2/2026

CHEEM: Continual Learning by Reuse, New, Adapt and Skip -- A Hierarchical Exploration-Exploitation Approach

Continual learning framework using hierarchical exploration-exploitation to acquire knowledge from task streams without catastrophic forgetting.

Ax Anders Sj\"oberg, Jakob Lindqvist, Magnus \"Onnheim, Mats Jirstrand, Lennart Svensson 4/2/2026

MCMC-Correction of Score-Based Diffusion Models for Model Composition

Combines MCMC correction with score-based diffusion models using Metropolis-Hastings steps for improved sampling in model composition.

Ax Antonio Di Noia, Iuri Macocco, Aldo Glielmo, Alessandro Laio, Antonietta Mira 4/2/2026

Scale-adaptive and robust intrinsic dimension estimation via optimal neighbourhood identification

Method for estimating intrinsic dimensionality of datasets accounting for scale-dependent effects and measurement noise in unsupervised learning.

Ax Bhrij Patel, Souradip Chakraborty, Mengdi Wang, Dinesh Manocha, Amrit Singh Bedi 4/2/2026

Code Comprehension then Auditing for Unsupervised LLM Evaluation

LLM-based approach for unsupervised code correctness evaluation that separates code comprehension from auditing to improve accuracy without reference implementations.

Ax Johnny Chan, Yuming Li 4/2/2026

Enhancing Team Diversity with Generative AI: A Novel Project Management Framework

Project management framework using GenAI agents to optimize team composition by matching personality roles.

Ax Chunyang Jiang, Chi-min Chan, Yiyang Cai, Yulong Liu, Wei Xue, Yike Guo 4/2/2026

Graceful Forgetting in Generative Language Models

Addresses negative transfer in fine-tuning by selectively forgetting unhelpful pre-trained knowledge in language models.

Ax Uranik Berisha, Jens Mehnert, Alexandru Paul Condurache 4/2/2026

Variance-Based Pruning for Accelerating and Compressing Trained Networks

Variance-based pruning method for compressing trained networks including Vision Transformers with minimal retraining.

Ax Xinfang Chen, Siyang Xiao, Xianying Zhu, Junhong Xie, Ming Liang, Dajun Chen, Wei Jiang, Yong Li, Peng Di 4/2/2026

NES: An Instruction-Free, Low-Latency Next Edit Suggestion Framework Powered by Learned Historical Editing Trajectories

NES framework for low-latency code edit suggestions without explicit instructions, using learned editing trajectories.

Ax A. Chervov, D. Fedoriaka, E. Konstantinova, A. Naumov, I. Kiselev, A. Sheveleva, I. Koltsov, S. Lytkin, A. Smolensky, A. Soibelman, F. Levkovich-Maslyuk, R. Grimov, D. Volovich, A. Isakov, A. Kostin, M. Litvinov, N. Vilkin-Krom, A. Bidzhiev, A. Krasnyi, M. Evseev, E. Geraseva, L. Grunwald, S. Galkin, E. Koldunov, S. Diner, A. Chevychelov, E. Kudasheva, A. Sychev, A. Kravchenko, Z. Kogan, A. Natyrova, L. Shishina, L. Cheldieva, V. Zamkovoy, D. Kovalenko, O. Papulov, S. Kudashev, D. Shiltsov, R. Turtayev, O. Nikitina, D. Mamayeva, S. Nikolenko, M. Obozov, A. Titarenko, A. Dolgorukova, A. Aparnev, O. Debeaupuis, S. Alami C., H. Isambert 4/2/2026

CayleyPy Growth: Efficient growth computations and hundreds of new conjectures on Cayley graphs (Brief version)

Open source CayleyPy library for efficient Cayley and Schreier graph computations, with 200+ new conjectures in group theory.

Ax Ammar Ahmed, Azal Ahmad Khan, Ayaan Ahmad, Sheng Di, Zirui Liu, Ali Anwar 4/2/2026

Retrieval-of-Thought: Efficient Reasoning via Reusing Thoughts

Retrieval-of-Thought (RoT) system reuses prior reasoning steps organized in thought graphs to improve LLM inference efficiency.

Ax Boxuan Zhang, Yi Yu, Jiaxuan Guo, Jing Shao 4/2/2026

Dive into the Agent Matrix: A Realistic Evaluation of Self-Replication Risk in LLM Agents

Evaluates self-replication risks in LLM agents through realistic testing of autonomous agent behaviors and safety concerns.

Ax Percy S. Zhai, So Won Jeong, Veronika Ro\v{c}kov\'a 4/2/2026

Conditional Flow Matching for Bayesian Posterior Inference

Proposes flow matching method for Bayesian posterior inference without likelihood evaluation, using block-triangular velocity fields.

Ax Mykolas Sveistrys, Richard Kunert 4/2/2026

PluriHopRAG: Exhaustive, Recall-Sensitive QA Through Corpus-Specific Document Structure Learning

RAG system for exhaustive multi-document question answering that checks all relevant documents without clear stopping conditions.

Ax Elias Hossain, Mehrdad Shoeibi, Ivan Garibay, Niloofar Yousefi 4/2/2026

BIOGEN: Evidence-Grounded Multi-Agent Reasoning Framework for Transcriptomic Interpretation in Antimicrobial Resistance

Multi-agent reasoning framework using AI agents for interpreting gene clusters in antimicrobial resistance transcriptomic data.

Ax Guneet S. Dhillon, Javier Gonz\'alez, Teodora Pandeva, Alicia Curth 4/2/2026

E-Scores for (In)Correctness Assessment of Generative Model Outputs

Framework using conformal prediction to assess correctness of LLM outputs and construct confidence sets for generative model responses.

Ax Wenlun Zhang, Yunshan Zhong, Zihao Ding, Xinyu Li, Kentaro Yoshioka 4/2/2026

D4C: Data-Free Quantization for Contrastive Language-Image Pre-training Models

Data-free quantization techniques for CLIP vision-language models enabling model compression without real data access for privacy-sensitive scenarios.

Ax Asad Aali, Muhammad Ahmed Mohsin, Vasiliki Bikia, Arnav Singhvi, Richard Gaus, Suhana Bedi, Hejie Cui, Miguel Fuentes, Alyssa Unell, Yifan Mai, Jordan Cahoon, Michael Pfeffer, Roxana Daneshjou, Sanmi Koyejo, Emily Alsentzer, Christopher Potts, Nigam H. Shah, Akshay S. Chaudhari 4/2/2026