Isolater - Feed

Ax Cagri Eryilmaz 4/23/2026

ChipCraftBrain: Validation-First RTL Generation via Multi-Agent Orchestration

ChipCraftBrain: Multi-agent orchestration framework for LLM-based RTL code generation with validation-first approach and synthesis awareness.

Ax Venus Team, Sunhao Dai, Yong Deng, Jinzhen Lin, Yusheng Song, Guoqing Wang, Xiaofeng Wu, Yuqi Zhou, Shuo Yang, Zhenzhe Ying, Zhanwei Zhang, Changhua Meng, Weiqiang Wang 4/23/2026

DR-Venus: Towards Frontier Edge-Scale Deep Research Agents with Only 10K Open Data

DR-Venus: 4B edge-scale deep research agent trained on 10K open data for cost-effective autonomous research task execution.

Ax Chenxi Zhou, Pengfei Cao, Jiang Li, Bohan Yu, Jinyu Ye, Jun Zhao, Kang Liu 4/23/2026

From Signal Degradation to Computation Collapse: Uncovering the Two Failure Modes of LLM Quantization

Mechanistic analysis of LLM quantization failure modes, revealing distinct mechanisms between 4-bit and 2-bit precision degradation.

Ax Giorgia Gulino, Manuel Petrucci 4/23/2026

Depression Risk Assessment in Social Media via Large Language Models

LLM-based system for depression risk assessment in Reddit posts using multi-label classification on linguistic signals.

Ax Zijie Li, Yichun Shi, Jingxiang Sun, Ye Wang, Yixuan Huang, Zhiyao Guo, Xiaochen Lian, Peihao Zhu, Yu Tian, Zhonghua Zhai, Peng Wang 4/23/2026

MMCORE: MultiModal COnnection with Representation Aligned Latent Embeddings

MMCORE: Unified multimodal framework using vision-language models to generate semantic embeddings for image diffusion models.

Ax Sheikh Junaid Fayaz, Nestor D. Montiel-Bohorquez, Wilson Ricardo Leal da Silva, Shashank Bishnoi, Matteo Romano, Manuele Gatti, N. M. Anoop Krishnan 4/23/2026

A Multi-Plant Machine Learning Framework for Emission Prediction, Forecasting, and Control in Cement Manufacturing

Data-driven ML framework for emission prediction and control in cement manufacturing using operational data from 4 plants.

Ax Shilei Luo, Zhiqi Zhang, Hengchen Dai, Dennis Zhang 4/23/2026

Behavioral Transfer in AI Agents: Evidence and Privacy Implications

Study examining whether LLM-powered agents systematically reflect behavioral characteristics of their human operators.

Ax Fateme Rahmani, Mahdi Jafari Siavoshani, Mohammad Hossein Rohban 4/23/2026

Generalization and Membership Inference Attack a Practical Perspective

Empirical study reassessing correlation between membership inference attack success rates and model generalization.

Ax Palawat Busaranuvong, Reza Saadati Fard, Emmanuel Agu, Deepak Kumar, Shefalika Gautam, Bengisu Tulu, Diane Strong 4/23/2026

Infection-Reasoner: A Compact Vision-Language Model for Wound Infection Classification with Evidence-Grounded Clinical Reasoning

4B-parameter vision-language model for wound infection classification with evidence-grounded clinical reasoning explanations.

Ax Divyanshu Goyal, Akhil Eppa, Vanya Bannihatti Kumar 4/23/2026

DistortBench: Benchmarking Vision Language Models on Image Distortion Identification

DistortBench: Diagnostic benchmark evaluating vision-language models on image distortion identification across 27 types.

Ax Xuxin Tang, Ibrahim Tahmid, Eric Krokos, Kirsten Whitley, Xuan Wang, Chris North 4/23/2026

Semantic Prompting: Agentic Incremental Narrative Refinement through Spatial Semantic Interaction

Semantic prompting: Interactive approach enabling agentic LLM-based narrative refinement through spatial semantic interactions.

Ax Huy Nghiem, Phuong-Anh Nguyen-Le, Sy-Tuyen Ho, Hal Daume III 4/23/2026

Bias in the Tails: How Name-conditioned Evaluative Framing in Resume Summaries Destabilizes LLM-based Hiring

Analysis of name-based bias in LLM-generated resume summaries across 4 models using synthetic and real-world data.

Ax Yiyang Du, Zhanqiu Guo, Xin Ye, Liu Ren, Chenyan Xiong 4/23/2026

EmbodiedMidtrain: Bridging the Gap between Vision-Language Models and Vision-Language-Action Models via Mid-training

EmbodiedMidtrain bridges gap between vision-language models and embodied vision-language-action models through mid-training adaptation.

Ax Yusuf Kesmen, Fay Elhassan, Jiayi Ma, Julien Stalhandske, David Sasu, Alexandra Kulinkina, Akhil Arora, Lars Klein, Mary-Anne Hartley 4/23/2026

Statistics, Not Scale: Modular Medical Dialogue with Bayesian Belief Engine

BMBE: Modular medical dialogue framework separating language models from probabilistic reasoning for autonomous diagnostic agents.

Ax Ethan Knights 4/23/2026

Cognitive Alignment At No Cost: Inducing Human Attention Biases For Interpretable Vision Transformers

Fine-tuning Vision Transformers' self-attention on human saliency maps to align model attention with human visual perception patterns.

Ax Ziyi Wang, Chen Zhang, Wenjun Peng, Qi Wu, Xinyu Wang 4/23/2026

TriEx: A Game-based Tri-View Framework for Explaining Internal Reasoning in Multi-Agent LLMs

TriEx framework explaining multi-agent LLM reasoning through structured first-person self-reasoning and belief state visualization.

Ax Spyros Galanis 4/23/2026

Information Aggregation with AI Agents

Empirical study of LLM agents aggregating private information through trading in prediction markets.

Ax Sadra Sabouri, Zeinabsadat Saghi, Run Huang, Sujay Maladi, Esmeralda Eufracio, Sumit Gulwani, Souti Chattopadhyay 4/23/2026

Auditing and Controlling AI Agent Actions in Spreadsheets

System for auditing and controlling AI agent actions in spreadsheets during execution with user oversight.

Ax Yicheng Pan, Ruisong Zhou, Haijun Zou, Tianyou Li, Zaiwen Wen 4/23/2026

Learning to Solve the Quadratic Assignment Problem with Warm-Started MCMC Finetuning

PLMA framework combining learning with warm-started MCMC for solving quadratic assignment problems.

Ax Xuelin Zhang, Peipei Yuan 4/23/2026

On the Stability and Generalization of First-order Bilevel Minimax Optimization

Theoretical analysis of bilevel minimax optimization algorithms' generalization properties for hyperparameter optimization and reinforcement learning.

Ax Natalia Martinez Gil, Fearghal O'Donncha, Wesley M. Gifford, Nianjun Zhou, Dhaval C. Patel, Roman Vaculin 4/23/2026

Adaptive Conformal Anomaly Detection with Time Series Foundation Models for Signal Monitoring

Adaptive conformal anomaly detection method using pre-trained time series foundation models for signal monitoring with interpretable p-value anomaly scores.

Ax Joyjit Roy, Samaresh Kumar Singh 4/23/2026

AgentSOC: A Multi-Layer Agentic AI Framework for Security Operations Automation

AgentSOC multi-layer agentic framework for SOC automation integrating perception, anticipatory reasoning, and risk-based response action selection for security alerts.

Ax Weitong Kong, Di Wen, Kunyu Peng, David Schneider, Zeyun Zhong, Alexander Jaus, Zdravko Marinov, Jiale Wei, Ruiping Liu, Junwei Zheng, Yufan Chen, Lei Qi, Rainer Stiefelhagen 4/23/2026

IMPACT-CYCLE: A Contract-Based Multi-Agent System for Claim-Level Supervisory Correction of Long-Video Semantic Memory

IMPACT-CYCLE multi-agent contract-based system for claim-level supervisory correction of long-video semantic memory with interpretable intermediate outputs.

Ax Sachin Kumar 4/23/2026

Meta-Tool: Efficient Few-Shot Tool Adaptation for Small Language Models

Meta-Tool empirical study comparing hypernetwork-based LoRA adaptation vs few-shot prompting for tool-use in 3B small language models.

Ax Ronghao Ni, Mihai Christodorescu, Limin Jia 4/23/2026

Taint-Style Vulnerability Detection and Confirmation for Node.js Packages Using LLM Agent Reasoning

LLM-agent-based vulnerability detection system for Node.js packages using multi-step reasoning to overcome dynamic JavaScript analysis limitations.

Ax Zehong Ke, Yanbo Jiang, Jinhao Li, Zhiyuan Liu, Yiqian Tu, Qingwen Meng, Heye Huang, Jianqiang Wang 4/23/2026

From Scene to Object: Text-Guided Dual-Gaze Prediction

Text-guided dual-gaze prediction model for interpretable driver attention using Vision-Language Models with scene-to-object semantic reasoning.

Ax He Yang Yuan, Xin Wang, Kundi Yao, An Ran Chen, Zishuo Ding, Zhenhao Li 4/23/2026

Towards Secure Logging: Characterizing and Benchmarking Logging Code Security Issues with LLMs

Characterization and benchmarking of logging code security vulnerabilities using LLMs to detect insecure logging practices and sensitive information exposure.

Ax Wenhong Zhu, Ruobing Xie, Rui Wang, Pengfei Liu 4/23/2026

Hybrid Policy Distillation for LLMs

Hybrid Policy Distillation framework unifying knowledge distillation methods for LLM compression, reformulating KD as reweighted token-level log-likelihood objective.

Ax Adriana Aida, Walida Amer, Katarina Bankovic, Dhruv Behl, Fabian Busch, Annie Bhalla, Minh Duong, Florian Gienger, Rohan Godse, Denis Grachev, Ralf Gulde, Elisa Hagensieker, Junpeng Hu, Shivam Joshi, Tobias Knoblauch, Likith Kumar, Damien LaRocque, Keerthana Lokesh, Omar Moured, Khiem Nguyen, Christian Preyss, Ranjith Sriganesan, Vikram Singh, Carsten Sponner, Anh Tong, Dominik Tuscher, Marc Tuscher, Pavan Upputuri 4/23/2026

Cortex 2.0: Grounding World Models in Real-World Industrial Deployment

Cortex 2.0 robotic system shifting from reactive Vision-Language-Action models to proactive world models for long-horizon manipulation tasks across embodiments.

Ax Jianxin Gao, Ruohan Lei, Wanli Peng 4/23/2026

Text Steganography with Dynamic Codebook and Multimodal Large Language Model

Text steganography method using LLMs with dynamic codebook and multimodal models for hiding information in generated text.

Ax Jeonghyeon Kim, Byeongjun Joung, Junwon Lee, Joohyung Lee, Taehoon Min, Sunjae Lee 4/23/2026

AgentLens: Adaptive Visual Modalities for Human-Agent Interaction in Mobile GUI Agents

Mobile GUI agent framework with adaptive visual modality selection for transparent human-agent interaction during smartphone task automation.

Ax Ryo Tamura, Haruhiko Morito, Yuna Oikawa, Guillaume Deffrennes, Shoichi Matsuda, Naruki Yoshikawa, Tomoaki Takayama, Taichi Abe, Koji Tsuda, Kei Terayama 4/23/2026

LLM-guided phase diagram construction through high-throughput experimentation

LLM used as experimental planner in closed-loop system for phase diagram construction of multicomponent alloys via high-throughput synthesis and X-ray diffraction.

Ax Xiang Shi, Shuaizhi Cheng, Mingwei Li 4/23/2026

Formalising the Logit Shift Induced by LoRA: A Technical Note

Technical formalization of logit shifts induced by LoRA fine-tuning, decomposing multi-layer effects into layerwise contributions using Fréchet approximation.

Ax Valentin Gabeur, Shangbang Long, Songyou Peng, Paul Voigtlaender, Shuyang Sun, Yanan Bao, Karen Truong, Zhicheng Wang, Wenlei Zhou, Jonathan T. Barron, Kyle Genova, Nithish Kannen, Sherry Ben, Yandong Li, Mandy Guo, Suhas Yogin, Yiming Gu, Huizhong Chen, Oliver Wang, Saining Xie, Howard Zhou, Kaiming He, Thomas Funkhouser, Jean-Baptiste Alayrac, Radu Soricut 4/23/2026

Image Generators are Generalist Vision Learners

Demonstrates that image generators develop strong zero-shot visual understanding capabilities through generative pretraining, similar to emergent LLM abilities.

Ax Changho Han (Medical Big Data Research Center, Seoul National University Medical Research Center, Seoul National University College of Medicine, Seoul, Republic of Korea), Songsoo Kim (Department of Biomedical Systems Informatics, Yonsei University College of Medicine, Seoul, Republic of Korea), Dong Won Kim (Department of Biomedical Systems Informatics, Yonsei University College of Medicine, Seoul, Republic of Korea), Leo Anthony Celi (Laboratory for Computational Physiology, Massachusetts Institute of Technology, Cambridge, MA, USA, Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA), Jaewoong Kim (Department of Biomedical Systems Informatics, Yonsei University College of Medicine, Seoul, Republic of Korea), SungA Bae (Department of Cardiology, Yongin Severance Hospital, Yonsei University College of Medicine, Yongin, Republic of Korea, Center for Digital Health, Yongin Severance Hospital, Yonsei University Health System, Yongin, Republic of Korea), Dukyong Yoon (Department of Biomedical Systems Informatics, Yonsei University College of Medicine, Seoul, Republic of Korea, Institute for Innovation in Digital Healthcare, Severance Hospital, Seoul, Republic of Korea) 4/23/2026

Surrogate modeling for interpreting black-box LLMs in medical predictions

Surrogate modeling framework to interpret and explain knowledge encoded in black-box LLMs for medical prediction tasks.

Ax Yuelin Zhang, Qingpeng Ding, Longxiang Tang, Chengyu Fang, Shing Shin Cheng 4/23/2026

A Vision-Language-Action Model for Adaptive Ultrasound-Guided Needle Insertion and Needle Tracking

Vision-Language-Action model for automated ultrasound-guided needle insertion, replacing hand-crafted pipelines with learned adaptive control.

Ax Alessio Palma, Indro Spinelli, Vignesh Prasad, Luca Scofano, Yufeng Jin, Georgia Chalvatzaki, Fabio Galasso 4/23/2026

Bimanual Robot Manipulation via Multi-Agent In-Context Learning

Using in-context learning with LLMs to enable bimanual robot manipulation without task-specific training, addressing coordination constraints in high-dimensional action spaces.

Ax Zhe Feng, Sen Lian, Changwei Wang, Muyang Zhang, Tianlong Tan, Rongtao Xu, Weiliang Meng, Xiaopeng Zhang 4/23/2026

LaplacianFormer:Rethinking Linear Attention with Laplacian Kernel

LaplacianFormer proposes Transformer variant using Laplacian kernel to reduce attention complexity for high-resolution vision tasks.

Ax Ramdhan Wibawa, Birendra Jha 4/23/2026

AI models of unstable flow exhibit hallucination

Research identifies hallucination in AI fluid dynamics models, where viscous fingering patterns appear realistic but are physically implausible.

Ax Gustav Keppler, Ghada Elbez, Veit Hagenmeyer 4/23/2026

CyberCertBench: Evaluating LLMs in Cybersecurity Certification Knowledge

CyberCertBench introduces benchmarks evaluating LLM domain knowledge against cybersecurity industry certification standards.

Ax Deevashwer Rathee, Jean-Luc Watson, Zirui Neil Zhao, G. Edward Suh, Raluca Ada Popa 4/23/2026

Onyx: Cost-Efficient Disk-Oblivious ANN Search

Research paper on Onyx, a cost-efficient disk-based approximate nearest neighbor search for sensitive data using oblivious RAM techniques.

Ax Leonardo Kuffo, Ioanna Tsakalidou, Roberta De Viti, Albert Angel, Ji\v{r}\'i I\v{s}a, Rastislav Lenhardt 4/23/2026

Semantic Recall for Vector Search

Paper introduces Semantic Recall metric for evaluating approximate nearest neighbor search by filtering semantically irrelevant objects.

Ax Hung Cuong Pham, Fatih Gedikli 4/23/2026

Scalable AI Inference: Performance Analysis and Optimization of AI Model Serving

Performance analysis and optimization of BentoML-based AI inference system for scalable model serving and deployment.

Ax Petrus Lipsanen, Liisa Rannikko, Fran\c{c}ois Christophe, Konsta Kalliokoski, Vlad Stirbu, Tommi Mikkonen 4/23/2026

Shift-Up: A Framework for Software Engineering Guardrails in AI-native Software Development -- Initial Findings

Shift-Up: Framework establishing software engineering guardrails for AI-native development, addressing architectural drift and maintainability.

Ax Neemesh Yadav, Palakorn Achananuparp, Jing Jiang, Ee-Peng Lim 4/23/2026

DialToM: A Theory of Mind Benchmark for Forecasting State-Driven Dialogue Trajectories

DialToM: Benchmark evaluating LLM Theory of Mind abilities through dialogue state prediction and functional utility assessment.

Ax Qianxi Hua, Xinyue Li, Zheng Yan, Yang Li, Chi Zhang, Yongyao Li, Yufei Liu 4/23/2026

VTouch++: A Multimodal Dataset with Vision-Based Tactile Enhancement for Bimanual Manipulation

VTouch++: Bimanual robot manipulation dataset with vision-based tactile sensing for contact-rich tasks.

Ax Markus Knauer, Edoardo Fiorini, Maximilian M\"uhlbauer, Stefan Schneyer, Promwat Angsuratanawech, Florian Samuel Lay, Timo Bachmann, Samuel Bustamante, Korbinian Nottensteiner, Freek Stulp, Alin Albu-Sch\"affer, Jo\~ao Silv\'erio, Thomas Eiband 4/23/2026