Ax Haohui Huang, Xi Yuan, Panpan Liao, Tao Teng, Chenguang Yang, Jing Guo, Yi Guo 16d ago

GenVid2Robot: From Video Generation to Robot Manipulation via Rigid-Geometric Consistency

Framework converting generated videos to robot manipulation trajectories by enforcing rigid-geometric consistency and kinematic constraints.

Ax Konstantin Garbers, Nicholas Oh 16d ago

Complexity-Guided Component-wise Initialization for Language Model Pretraining

Method for initializing language model pretraining using spectral patterns from pretrained checkpoints to improve efficiency.

Ax Rahul Jaiswal 16d ago

Leveraging Interpretable Tsetlin Machine for PDF Malware Detection

arXiv paper presenting Tsetlin Machine approach for interpretable PDF malware detection addressing cyberattack vector.

Ax Isarun Chamveha, Tretap Promwiset, Napat Wanchaitanawong, Trongtum Tongdee, Pairash Saiviroonporn, Warasinee Chaisangmongkon 16d ago

From Classification to Localization and Clinical Validation: Large-Scale Development of a Deep Learning System for Thoracic Disease Detection on Chest Radiographs in Thailand

arXiv paper on deep learning system for thoracic disease detection in chest radiographs developed and validated on Thai population data.

Ax Kwan Soo Shin, In Seok Kang, Yunkyung Min 16d ago

Creativity, honesty and designed forgetting emerge in small hyperbolic language models

arXiv paper studying creativity, honesty and memory in small hyperbolic language models and implications for AI assistants becoming companions.

Ax Miguel Arana-Catania, Gillian Pink, Glenn Roe 16d ago

Automatic Thematic Indexing of Large Literary Corpora: A Machine Learning Approach to Voltaire's Complete Works

arXiv paper applying machine learning to automatic thematic indexing of Voltaire's complete works, automating scholarly text categorization at scale.

Ax Miguel Arana-Catania, Catherine Conisbee, Matthew Kidd 16d ago

Letting the Data Speak: Extracting Keywords from Crowdsourced Collections with AI

arXiv paper evaluating NLP approaches for automated keyword extraction from crowdsourced Second World War digital collections.

Ax Cedric Caruzzo, Donggeun Yoo, Tae Soo Kim 16d ago

Deceptive Grounding: Entity Attribution Failure in Clinical Retrieval-Augmented Generation

arXiv paper identifying entity attribution failure in clinical RAG systems where retrieved evidence can be misattributed despite passing standard evaluation metrics.

Ax The Soofi-Team, :, Benedikt Droste, David Fitzek, Ruben H\"arle, Lukas Helff, Maximilian Idahl, Alex Jude, Abbas Goher Khan, Maurice Kraus, Timm Ruland, Richard Rutmann, Sebastian Sztwiertnia, Markus Frey, Daniil Gurgurov, Jan Pfister, Tom R\"ohr, Sebastian von Rohrscheidt, J\"org Bienert, Nicolas Flores-Herr, Simon Gottschalk, Andreas Hotho, Kristian Kersting, Joachim K\"ohler, Alexander L\"oser, Wolfgang Nejdl, Simon Ostermann, Jan Plogsties, Patrick Putzky, Mehdi Ali, Michael Fromm, Max L\"ubbering 16d ago

A Sovereign, Open-Source Foundation Model for German and English

Soofi S 30B-A3B: open-source MoE hybrid Mamba Transformer foundation model for German and English, 3B active parameters per token, optimized for long-context deployment.

Ax Tam\'as Matuszka, Andr\'as Tam\'asy, Bal\'azs Szol\'ar 16d ago

Multimodal Scenario Similarity Search for Autonomous Driving

Multimodal framework for retrieving similar autonomous driving scenarios combining visual and motion-based representations.

Ax Spiros Baxevanakis, Peng-Jian Yang 16d ago

Test-Time Scaling for Small VLMs on Multilingual Visual MCQ

Evaluates test-time scaling techniques on multilingual visual multiple-choice benchmark comparing self-consistency and beam search strategies on small open VLMs.

Ax Xingyu Zhu, Huanshen Wu, Shuo Wang, Beier Zhu, Jiannan Ge, Jiaheng Zhang, Long Chen 16d ago

Robustifying Vision-Language Models via Test-Time Prompt Adaptation

Test-time prompt adaptation method for improving Vision-Language Model robustness against adversarial perturbations using distributional structure awareness.

Ax Ibrahim Batuhan Akkaya, Kishaan Jeeveswaran, Bahram Zonooz, Elahe Arani 16d ago

Foveation-Guided Dynamic Token Selection for Robust and Efficient Vision Transformers

Foveated Dynamic Transformer applies human visual system principles for efficient vision transformer token selection with robustness to adversarial perturbations.

Ax Ahmed Oumar El-Shangiti, Abzal Nurgazy, Hilal AlQuabeh, Nikolai Rozanov, Kentaro Inui 16d ago

The Count Is There, but Misaligned: Understanding and Correcting Counting Failures in VLMs

Investigation of counting failures in VLMs reveals internal representations encode count information but fail verbalization; proposes probing-based detection method.

Ax Yuan Cao, Haiqian Yang 16d ago

Beyond Fixed Representations: The Vocabulary and Verifier Gaps in Open-Ended AI

Analysis of representational limitations in AI systems for reasoning, coding, theorem-proving and tool use; identifies vocabulary and verifier gaps in open-ended evaluation.

Ax Okezzi Ukorigho, Opeoluwa Owoyele 16d ago

Entropy-Constrained Machine Learning with Residual Data Augmentation for Modeling Chemical Kinetics

Physics-constrained ML framework using residual data augmentation and entropy constraints for accelerating turbulent flow simulations via surrogate chemical kinetics models.

Ax Kangwei Xu, Bing Li, Ulf Schlichtmann 16d ago

LLM for EDA in Front-End Design: Challenges and Opportunities

ArXiv paper on LLM applications for electronic design automation including HDL generation, testbench construction, and design space exploration.

Ax Federico L. Perlino, Oliver Hamelijnck, Adam M. Johansen, Theodoros Damoulas 16d ago

Deep Gaussian Processes on Directed Acyclic Graphs

ArXiv paper on deep Gaussian processes over DAGs for modeling compositions of partially-observed functions in causal and engineering systems.

Ax Kexin Li, Aastha Mehta, David Lie 16d ago

LDPKiT: Superimposing Remote Queries for Privacy-Preserving Distillation

Privacy-preserving model distillation framework using local differential privacy for remote API access scenarios. Non-adversarial approach.

Ax Takuya Kanayama, Yuki Ito, Tomoyuki Tamura, Masayuki Karasuyama 16d ago

Information-Theoretic Bayesian Optimization for Bilevel Optimization Problems

Bayesian optimization method for bilevel optimization with expensive black-box functions. Theoretical framework without clear AI/ML application focus.

Ax Liangyu Ding, Chenghan Wu, Guokai Li, Zizhuo Wang 16d ago

Self-Improving Neural-Guided Pruning: A Graph Neural Network Framework for Scalable Mixed Bundle Pricing

Graph neural network for mixed bundle pricing with self-improving pruning strategy.

Ax Willa Potosnak, Malcolm Wolff, Mengfei Cao, Ruijun Ma, Tatiana Konstantinova, Dmitry Efimov, Michael W. Mahoney, Boris Oreshkin, Kin G. Olivares 16d ago

Forking-Sequences: Statistically and Computationally Efficient Multi-Horizon Forecasting with Reduced Volatility

Multi-horizon time series forecasting with reduced volatility using neural sequence forking.

Ax Yavuz Bakman, Duygu Nur Yaldiz, Eleni Triantafillou, Peter Kairouz, Salman Avestimehr, Sai Praneeth Karimireddy 16d ago

Hair-Trigger Alignment: Black-Box Evaluation Cannot Guarantee Post-Update Alignment

Analysis of LLM alignment degradation after fine-tuning beyond black-box evaluation limits.

Ax Tanmay Ambadkar, Sourav Panda, Shreyash Kale, Jonathan Dodge, Abhinav Verma 16d ago

Preference Conditioned Multi-Objective Reinforcement Learning: Decomposed, Diversity-Driven Policy Optimization

Multi-objective RL approach for preference-conditioned policy optimization with diversity.

Ax Shaan Pakala, Aldair E. Gongora, Brian Giera, Evangelos E. Papalexakis 16d ago

Tensor Methods: A Unified and Interpretable Approach for Material Design

Tensor methods for material design optimization reducing computational search space.

Ax Yiyang Lu, Haresh Jadav, Mohammad Pedramfar, Ranveer Singh, Vaneet Aggarwal 16d ago

Upper-Linearizability of Online Non-Monotone DR-Submodular Maximization over Down-Closed Convex Sets

Online optimization of non-monotone DR-submodular functions with linearizability analysis.

Ax Ravi Ghadia, Maksim Abraham, Sergei Vorobyov, Max Ryabinin 16d ago

Untied Ulysses: Memory-Efficient Context Parallelism via Headwise Chunking

Memory-efficient context parallelism for long-sequence Transformer processing via headwise chunking.

Ax Luca Schmidt, Nina Effenberger, Vitus Benson, Philine L. Bommer, Robert Brunstein, Mikel N. Legasa, Maxim Samarin, Maybritt Schillinger 16d ago

Bridging the Gap Between Climate Science and Machine Learning in Climate Model Emulation

ML emulators for climate model acceleration to reduce computational costs in climate science.

Ax Orhun Bugra Baran, Melih Kandemir, Ramazan Gokberk Cinbis 16d ago

Policy-based Tuning of Autoregressive Image Models with Instance- and Distribution-Level Rewards

RL method for tuning autoregressive image models with instance and distribution-level rewards.

Ax Haishan Ye 16d ago

Logarithmic High-Probability Regret for Online Convex Optimization with Two-Point Bandit Feedback

Research paper on logarithmic regret bounds for online convex optimization with bandit feedback.

Ax Hannah Guan, Soukayna Mouatadid, Paulo Orenstein, Judah Cohen, Haiyu Dong, Zekun Ni, Jeremy Berman, Genevieve Flaspohler, Alex Lu, Jakob Schloer, Joshua Talib, Jonathan A. Weyn, Lester Mackey 16d ago

Enhancing AI and Dynamical Subseasonal Forecasts with Probabilistic Bias Correction

Probabilistic bias correction improves AI and physics-based subseasonal weather forecasts.

Ax Tao Xie, Zexi Tan, Haoyi Xiao, Mengke Li, Yiqun Zhang, Yang Lu, Cuie Yang, Yiu-ming Cheung 16d ago

AnchorMoE: Interpretable Time Series Classification via Anchor-Routed MoE

Interpretable multivariate time series classification via anchor-routed MoE for high-stakes applications.

Ax Ali Behrouz, Farnoosh Hashemi, Adel Javanmard, Vahab Mirrokni 16d ago

Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories

Method for LLMs to self-modify and consolidate in-context knowledge into long-term parameters through memory consolidation.

Ax Omar Adalat, Edwin Hamel-De le Court, Francesco Belardinelli 16d ago

Contract-Based Compositional Shielding for Safe Multi-Agent Reinforcement Learning

Contract-based compositional shielding enforces safety in decentralized multi-agent RL through runtime guarantees.

Ax Carlo Di Cicco 16d ago

Code Correctness Signals in LLM Hidden States: Pre-Generation Probing and Repair Geometry

Code correctness is linearly decodable from LLM hidden states before generation; studies repair geometry in Qwen3.

Ax Alan Nadelsticher Ruvalcaba 16d ago

Evolutionary Discovery of Developmental Reward Schedules in Deep Reinforcement Learning

Evolutionary framework discovers developmental reward schedules in RL combining agency, novelty, and reactivity components.

Ax Amirhossein Kardoost, Lion Gleiter, Tingying Peng, Carsten Marr 16d ago

3D Masked Autoencoders are Robust Learners of Volumetric and Multimodal Cellular Representations for Microscopy

3D masked autoencoders outperform 2D variants for volumetric microscopy cell representations in self-supervised learning.

Ax Yujiang He, Frederic Uhrweiller, Bernhard Sick 16d ago

Towards Continuous Power Forecasting: Practical Continual Learning for Real-World Energy Systems in Nonstationary Time Series

Continual learning approach for power forecasting in nonstationary energy systems using deep learning.

Ax Aniq Ur Rahman 16d ago

Estimation-Prediction Tradeoff in Causal Probabilistic Temporal Graphs

Temporal link prediction tradeoff analysis: Fisher information bounds on parameter estimation vs predictive accuracy in causal probabilistic temporal graphs.

Ax Zijun Xie, Binbin Zheng, Enlei Gong, Jihua Liu, Yuyang You, Lingfeng Liu, Jiayao Tang, Guanqun Zhao, Aoqi Hu, Zeyu Chen 16d ago

ECHO: Prune To Act, Trace To Learn With Selective Turn Memory In Agentic RL

ECHO: memory-efficient context management for long-horizon language agents via selective turn pruning and tracing; maintains interpretability of decision support.

Ax Rajat Ghosh 16d ago

Reduced-Order Models: The Mother of World Models

World models as functional descendants of model-order-reduction techniques; argues modern self-supervised approaches reimplement decades-old control theory concepts.

Ax Zhang Gongyue, Wang Zhiyong, Liu Donghan, Ren Weihong, Sheng Yixuan, Liu Honghai 16d ago

The Anatomy of Implicit Bias: Information Allocation in Neural Network Training

Training-time information allocation view of implicit bias in neural networks; explains how optimization forms writing patterns for error signals during training.

Ax Adrian Ciotinga, Yeming Dai, YooJung Choi 16d ago

PeTeR: Post-Training Robustification of Probabilistic Circuits

PeTeR: post-training robustification of probabilistic circuits via distributionally-robust optimization to improve generalization under noise and distribution shift.

Ax Zhoujie Hou, Song Wang, Kexin Lou, Mo Wang, Chen Wei, Quanying Liu 16d ago

Omni-Sleep: A Sleep Foundation Model via Hierarchical Contrastive Learning of CNS-ANS Dynamics

Omni-Sleep: foundation model for sleep staging using hierarchical contrastive learning on multimodal polysomnography signals (EEG, ECG, respiration).

Ax Haozhan Tang, Zerui Wang, Yuxian Gu, Song Han, Han Cai 16d ago

Jet-Long: Efficient Long-Context Extension with Dynamic Bifocal RoPE

Jet-Long: efficient long-context extension for LLMs using dynamic bifocal RoPE, enables zero-shot context extension 10x beyond pretraining window.

Ax Teng-Ruei Chen 16d ago

Resample or Reroute? Budget-Aware Test-Time Model Selection for Large Language Models

Budget-aware test-time selection for routing LLM queries across multiple models; resampling strategy improves cost-quality tradeoff vs single-router approach.

Ax Somnath Chakraborty, Hariharan Narayanan 16d ago

A Fourier analytique approach to Gaussian mixture learning

Fourier analytic method for learning Gaussian mixture models with theoretical guarantees on separating mixture components under distance constraints.

Ax Xiang Cheng, Sangdon Park, HyungSeok Han, Xiaokuan Zhang, Taesoo Kim 16d ago

Ruby: Unmasking Unsafe Rust in Stripped Binaries via Machine Learning

Ruby: ML-based tool to detect unsafe Rust regions in stripped binaries without source code access; addresses memory safety analysis gap.

Ax Tianxi Cai, Feiqing Huang, Ryumei Nakada, Linjun Zhang, Doudou Zhou 16d ago

Contrastive Learning on Multimodal Analysis of Electronic Health Records

Contrastive learning approach for multimodal EHR analysis combining structured codes and unstructured clinical notes via deep learning.

Ax Chris Junchi Li 16d ago

Accelerated Fully First-Order Methods for Bilevel and Minimax Optimization

arXiv paper on accelerated first-order methods for bilevel and minimax optimization.