Isolater - Feed

Ax Jiale Ding, Xiang Zheng, Yutao Wu, Cong Wang, Wei-Bin Lee, Ling Pan, Xingjun Ma, Yu-Gang Jiang 3/25/2026

RedTopic: Toward Topic-Diverse Red Teaming of Large Language Models

RedTopic framework for topic-diverse red teaming of LLMs to identify vulnerabilities across broad range of harmful topics adaptively.

Ax Said Ohamouddou, Hanaa El Afia, Mohamed Hamza Boulaich, Abdellatif El Afia, Raddouane Chiheb 3/25/2026

MS-DGCNN++: Multi-Scale Dynamic Graph Convolution with Scale-Dependent Normalization for Robust LiDAR Tree Species Classification

MS-DGCNN++ applies multi-scale dynamic graph convolution with scale-dependent normalization for LiDAR tree species classification.

Ax Muhao Guo, Jiaqi Wu, Yizheng Liao, Wenke Lee, Shengzhe Chen, Yang Weng 3/25/2026

Graph Structure Learning with Privacy Guarantees for Open Graph Data

Privacy-preserving graph structure learning framework for publishing open graph data with differential privacy guarantees.

Ax Mircea Lazar 3/25/2026

From Product Hilbert Spaces to the Generalized Koopman Operator and the Nonlinear Fundamental Lemma

Derives generalized Koopman operator solutions and nonlinear fundamental lemma for data-driven control of nonlinear systems.

Ax Yanzhou Li, Tianlin Li, Yiran Zhang, Shangqing Liu, Aishan Liu, Xianglong Liu, Yang Liu 3/25/2026

From Context to Intent: Reasoning-Guided Function-Level Code Completion

Reasoning-guided LLM function completion using context when docstrings are absent in real-world code repositories.

Ax Tao Wang, Zhenxuan Zhang, Yuanbo Zhou, Xinlin Zhang, Yuanbin Chen, Tao Tan, Guang Yang, Tong Tong 3/25/2026

From Noisy Labels to Intrinsic Structure: A Geometric-Structural Dual-Guided Framework for Noise-Robust Medical Image Segmentation

Proposes geometric-structural dual-guided framework for medical image segmentation robust to noisy labels.

Ax JiYuan Wang, Chunyu Lin, Lei Sun, Rongying Liu, Lang Nie, Mingxing Li, Kang Liao, Xiangxiang Chu 3/25/2026

From Editor to Dense Geometry Estimator

Analyzes fine-tuning image editing models versus text-to-image generators as foundations for dense geometry prediction tasks.

Ax Yi Yuan, Xubo Liu, Haohe Liu, Xiyuan Kang, Zhuo Chen, Yuxuan Wang, Mark D. Plumbley, Wenwu Wang 3/25/2026

DreamAudio: Customized Text-to-Audio Generation with Diffusion Models

DreamAudio enables fine-grained control over acoustic characteristics in text-to-audio generation using diffusion models.

Ax John Zheng, Farhad Maleki 3/25/2026

Selective Classifier-free Guidance for Zero-shot Text-to-speech

Applies classifier-free guidance from image generation to zero-shot text-to-speech synthesis for balancing speaker fidelity and text adherence.

Ax Xiao Wang, Jia Wang, Yijie Wang, Pengtao Dang, Sha Cao, Chi Zhang 3/25/2026

MARS: toward more efficient multi-agent collaboration for LLM reasoning

MARS proposes efficient multi-agent collaboration framework for LLM reasoning, reducing computational overhead of Multi-Agent Debate while maintaining reasoning capabilities.

Ax Mohamad Al Mdfaa, Svetlana Lukina, Timur Akhtyamov, Arthur Nigmatzyanov, Dmitrii Nalberskii, Sergey Zagoruyko, Gonzalo Ferrer 3/25/2026

VL-KnG: Persistent Spatiotemporal Knowledge Graphs from Egocentric Video for Embodied Scene Understanding

VL-KnG constructs spatiotemporal knowledge graphs from egocentric video using vision-language models for persistent scene understanding without 3D reconstruction.

Ax Nanaka Hosokawa, Ryo Takahashi, Tomoya Kitano, Yukihiro Iida, Chisako Muramatsu, Tatsuro Hayashi, Yuta Seino, Xiangrong Zhou, Takeshi Hara, Akitoshi Katsumata, Hiroshi Fujita 3/25/2026

Generating Findings for Jaw Cysts in Dental Panoramic Radiographs Using a GPT-Based VLM: A Preliminary Study on Building a Two-Stage Self-Correction Loop with Structured Output (SLSO) Framework

Self-correction Loop with Structured Output framework enhances GPT-based VLMs for generating reliable dental radiological findings in medical image interpretation.

Ax Fabio De Sousa Ribeiro, Ainkaran Santhirasekaram, Ben Glocker 3/25/2026

Counterfactual Identifiability via Dynamic Optimal Transport

Counterfactual identification framework using dynamic optimal transport addresses causal inference from observational data with high-dimensional multivariate outcomes.

Ax Haeji Jung, Jinju Kim, Kyungjin Kim, Youjeong Roh, David R. Mortensen 3/25/2026

Happiness is Sharing a Vocabulary: A Study of Transliteration Methods

Study investigates transliteration methods for bridging multilingual NLP gaps, examining shared scripts, vocabularies, and phonology in non-Latin languages.

Ax Guoqing Wang, Sunhao Dai, Guangze Ye, Zeyu Gan, Wei Yao, Yong Deng, Xiaofeng Wu, Zhenzhe Ying 3/25/2026

Information Gain-based Policy Optimization: A Simple and Effective Approach for Multi-Turn Search Agents

Information Gain-based Policy Optimization uses RL to train LLM agents for multi-turn search with tool use, addressing reward sparsity in exploration-based tasks.

Ax Dongsen Zhang, Zekun Li, Xu Luo, Xuannan Liu, Peipei Li, Wenjun Xu 3/25/2026

MCP Security Bench (MSB): Benchmarking Attacks Against Model Context Protocol in LLM Agents

MCP Security Bench systematically evaluates attacks against Model Context Protocol in LLM agents, measuring resistance of tool-calling systems to adversarial inputs.

Ax Sofiya Garkot, Maksym Shamrai, Ivan Synytsia, Mariya Hirna 3/25/2026

GUIrilla: A Scalable Framework for Automated Desktop UI Exploration

GUIrilla is a scalable framework for automated desktop UI exploration generating large-scale training data for LLM-based GUI understanding and automation.

Ax Fenfen Lin, Yesheng Liu, Haiyu Xu, Chen Yue, Zheqi He, Mingxuan Zhao, Miguel Hu Chen, Jiakang Liu, JG Yao, Xi Yang 3/25/2026

Do Vision-Language Models Measure Up? Benchmarking Visual Measurement Reading with MeasureBench

MeasureBench benchmarks vision-language models on visual measurement reading tasks with real-world and synthesized instrument images.

Ax Alina Fastowski, Bardh Prenkaj, Yuxiao Li, Gjergji Kasneci 3/25/2026

Injecting Falsehoods: Adversarial Man-in-the-Middle Attacks Undermining Factual Recall in LLMs

Xmera framework evaluates adversarial man-in-the-middle attacks on LLM factual recall through prompt injection, measuring vulnerability of question-answering systems.

Ax Zhanheng Nie, Chenghan Fu, Daoze Zhang, Junxian Wu, Wanxian Guan, Pengjie Wang, Jian Xu, Bo Zheng 3/25/2026

MOON2.0: Dynamic Modality-balanced Multimodal Representation Learning for E-commerce Product Understanding

MOON2.0 addresses multimodal imbalance in MLLMs for e-commerce product understanding through dynamic modality-balanced representation learning.

Ax Yuanzhe Li, Steffen M\"uller 3/25/2026

Pedestrian Crossing Intention Prediction Using Multimodal Fusion Network

Multimodal fusion network for pedestrian crossing intention prediction in autonomous vehicle systems.

Ax Jiajun Zhang, Shijia Luo, Ruikang Zhang, Qi Su 3/25/2026

HUMORCHAIN: Theory-Guided Multi-Stage Reasoning for Interpretable Multimodal Humor Generation

HumorChain: Theory-guided multi-stage reasoning framework for interpretable multimodal humor generation using LLMs.

Ax Yerim Jeon, Miso Lee, WonJun Moon, Jae-Pil Heo 3/25/2026

Masking Matters: Unlocking the Spatial Reasoning Capabilities of LLMs for 3D Scene-Language Understanding

Study on spatial reasoning in LLMs for 3D scene understanding, examining attention masking mechanisms for order-agnostic objects.

Ax Haicheng Liao, Huanming Shen, Bonan Wang, Yongkang Li, Yihong Tang, Chengyue Wang, Dingyi Zhuang, Kehua Chen, Hai Yang, Chengzhong Xu, Zhenning Li 3/25/2026

Think Before You Drive: World Model-Inspired Multimodal Grounding for Autonomous Vehicles

ThinkDeeper: Framework for autonomous vehicle grounding using world models for 3D spatial reasoning and scene prediction.

Ax Nikhil Verma, Joonas Linnosmaa, Leonardo Espinosa-Leal, Napat Vajragupta 3/25/2026

Arc Gradient Descent: A Geometrically Motivated Gradient Descent-based Optimiser with Phase-Aware, User-Controlled Step Dynamics (proof-of-concept)

ArcGD: Geometrically motivated gradient descent optimizer with phase-aware step dynamics, evaluated on benchmarks.

Ax Chenyu Zhang, Lanjun Wang, Yiwen Ma, Wenhui Li, Yi Tu, An-An Liu 3/25/2026

Metaphor-based Jailbreak Attacks on Text-to-Image Models

Research on metaphor-based jailbreak attacks against text-to-image models' safety defense mechanisms.

Ax Yu He, Da Huang, Zhenyang Liu, Zixiao Gu, Qiang Sun, Guangnan Ye, Yanwei Fu, Yu-Gang Jiang 3/25/2026

Schr\"odinger's Navigator: Imagining an Ensemble of Futures for Zero-Shot Object Navigation

Zero-shot object navigation for robots using ensemble prediction of future states in unseen, cluttered environments.

Ax Bhanu Prakash Vangala, Ali Adibifar, Ashish Gehani, Tanu Malik 3/25/2026

AI-Generated Code Is Not Reproducible (Yet): An Empirical Study of Dependency Gaps in LLM-Based Coding Agents

Empirical study examining reproducibility gaps in code generated by LLM coding agents and missing dependency specifications.

Ax Guanyuan Pan, Shuai Wang, Yugui Lin, Tiansheng Zhou, Pietro Li\`o, Zhenxin Zhao, Yaqi Wang 3/25/2026

VLM-CAD: VLM-Optimized Collaborative Agent Design Workflow for Analog Circuit Sizing

VLM-CAD: Collaborative agent design workflow for analog circuit sizing using vision-language models with spatial reasoning.

Ax Arjun Nichani (Richard), Hsiang Hsu (Richard), Chun-Fu (Richard), Chen, Haewon Jeong 3/25/2026

Does Privacy Always Harm Fairness? Data-Dependent Trade-offs via Chernoff Information Neural Estimation

Information-theoretic analysis of trade-offs between fairness, privacy, and accuracy in machine learning using Chernoff Information.

Ax Xinlei Yin, Xiulian Peng, Xiao Li, Zhiwei Xiong, Yan Lu 3/25/2026

Hierarchical Long Video Understanding with Audiovisual Entity Cohesion and Agentic Search

HAVEN: Framework for long-video understanding using agentic search and audiovisual entity cohesion to maintain global coherence.

Ax Zhiyu An, Wan Du 3/25/2026

Representational Homomorphism Predicts and Improves Compositional Generalization In Transformer Language Model

Analysis of representational homomorphism in transformers to predict and improve compositional generalization in language models.

Ax Wenxuan Huang, Yu Zeng, Qiuchen Wang, Zhen Fang, Shaosheng Cao, Zheng Chu, Qingyu Yin, Shuang Chen, Zhenfei Yin, Lin Chen, Zehui Chen, Xu Tang, Yao Hu, Shaohui Lin, Philip Torr, Feng Zhao, Wanli Ouyang 3/25/2026

Vision-DeepResearch: Incentivizing DeepResearch Capability in Multimodal Large Language Models

Vision-DeepResearch: Framework augmenting multimodal LLMs with tool-calling capabilities for visual and textual search.

Ax Yunwei Bai, Ying Kiat Tan, Yao Shu, Tsuhan Chen 3/25/2026

1S-DAug: One-Shot Data Augmentation for Robust Few-Shot Generalization

1S-DAug: One-shot data augmentation method for improved few-shot learning generalization using generative synthesis.

Ax Xinrong Chen, Xu Chu, Yingmin Qiu, Hengyuan Zhang, Jing Xiong, Shiyu Tang, Shuai Liu, Shaokang Yang, Cheng Yang, Hayden Kwok-Hay So, Ngai Wong 3/25/2026

Residual Decoding: Mitigating Hallucinations in Large Vision-Language Models via History-Aware Residual Guidance

Residual Decoding: Training method to reduce hallucinations in vision-language models using history-aware residual guidance.

Ax Hongwei Yan, Guanglong Sun, Kanglei Zhou, Qian Li, Liyuan Wang, Yi Zhong 3/25/2026

FlyPrompt: Brain-Inspired Random-Expanded Routing with Temporal-Ensemble Experts for General Continual Learning

FlyPrompt: Brain-inspired routing method for continual learning from non-stationary data streams without task boundaries.

Ax Zeping Li, Guancheng Wan, Keyang Chen, Yu Chen, Yiwen Zhao, Philip Torr, Guangnan Ye, Zhenfei Yin, Hongfeng Chai 3/25/2026

Behavioral Consistency Validation for LLM Agents: An Analysis of Trading-Style Switching through Stock-Market Simulation

Study evaluating behavioral consistency of LLM agents in stock market simulations against real market participant behavior.

Ax Xiaowen Tao, Yinuo Wang, Haitao Ding, Yuanyang Qi, Ziyu Song 3/25/2026

Energy-Aware Reinforcement Learning for Robotic Manipulation of Articulated Components in Infrastructure Operation and Maintenance

Energy-aware reinforcement learning for robotic manipulation of articulated objects in infrastructure maintenance and smart cities.

Ax Songming Zhang, Xue Zhang, Tong Zhang, Bojie Hu, Yufeng Chen, Jinan Xu 3/25/2026

KDFlow: A User-Friendly and Efficient Knowledge Distillation Framework for Large Language Models

KDFlow: Framework for efficient knowledge distillation of large language models into smaller models with heterogeneous training backends.

Ax Wenhao Wu, Zhentao Tang, Yafu Li, Shixiong Kai, Mingxuan Yuan, Chunlin Chen, Zhi Wang 3/25/2026

From Conflict to Consensus: Boosting Medical Reasoning via Multi-Round Agentic RAG

MA-RAG: Multi-round agentic RAG system for medical reasoning with LLMs, addressing hallucinations and outdated knowledge through iterative refinement.

Ax Kevin Vogt-Lowell, Theodoros Tsiligkaridis, Rodney Lafuente-Mercado, Surabhi Ghatti, Shanghua Gao, Marinka Zitnik, Daniela Rus 3/25/2026

When Sensors Fail: Temporal Sequence Models for Robust PPO under Sensor Drift

Augmenting Proximal Policy Optimization with temporal sequence models for robust reinforcement learning under sensor drift and partial observability.

Ax Jessica Sanson, Rahul C. Shah, Valerio Frascolla 3/25/2026

Human Presence Detection via Wi-Fi Range-Filtered Doppler Spectrum on Commodity Laptops

Wi-Fi based human presence detection using monostatic Doppler spectrum on commodity laptops without external sensors.

Ax Amos Goldman (NVIDIA Corporation), Nimrod Boker (NVIDIA Corporation), Maayan Sheraizin (NVIDIA Corporation), Nimrod Admoni (NVIDIA Corporation), Artem Polyakov (NVIDIA Corporation), Subhadeep Bhattacharya (NVIDIA Corporation), Fan Yu (NVIDIA Corporation), Kai Sun (NVIDIA Corporation), Georgios Theodorakis (NVIDIA Corporation), Hsin-Chun Yin (NVIDIA Corporation), Peter-Jan Gootzen (NVIDIA Corporation), Aamir Shafi (NVIDIA Corporation), Assaf Ravid (NVIDIA Corporation), Salvatore Di Girolamo (NVIDIA Corporation), Manjunath Gorentla Venkata (NVIDIA Corporation), Gil Bloch (NVIDIA Corporation) 3/25/2026