Ax Ghita Fassy El Fehri, Aur\'elien Bellet, Philippe Bastien 11d ago

Differentially Private and Federated Structure Learning in Bayesian Networks

Fed-Sparse-BNSL: Federated method for learning Bayesian network structures with differential privacy, addressing decentralized data challenges.

Ax Le Thien Phuc Nguyen, Zhuoran Yu, Samuel Low Yu Hang, Subin An, Jeongik Lee, Yohan Ban, SeungEun Chung, Thanh-Huy Nguyen, JuWan Maeng, Soochahn Lee, Yong Jae Lee 11d ago

See, Hear, and Understand: Benchmarking Audiovisual Human Speech Understanding in Multimodal Large Language Models

AV-SpeakerBench: Benchmark evaluating multimodal LLMs on fine-grained audiovisual speech understanding with 3,212 multiple-choice questions.

Ax Thao Nguyen, Sicheng Mo, Krishna Kumar Singh, Yilin Wang, Jing Shi, Nicholas Kolkin, Eli Shechtman, Yong Jae Lee, Yuheng Li 11d ago

Relational Visual Similarity

Research on relational visual similarity in AI vision systems, comparing current methods against human-like relational perception across different domains.

Ax Qiushi Han, David Simchi-Levi, Renfei Tan, Zishuo Zhao 11d ago

Multi-agent Adaptive Mechanism Design

DRAM: Framework combining mechanism design and online learning for sequential multi-agent settings to ensure truthful reporting with cost-optimality.

Ax Lee Hyoseok, Sohwi Lim, Eunju Cha, Tae-Hyun Oh 11d ago

Measurement-Consistent Langevin Corrector for Stabilizing Latent Diffusion Inverse Problem Solvers

Measurement-Consistent Langevin Corrector: Method stabilizing latent diffusion models for inverse problems by reducing discrepancy with learned reverse diffusion.

Ax Jacob Paul Simpson, Efstratios Palias, Sharu Theresa Jose 11d ago

Sample Complexity of Composite Quantum Hypothesis Testing

Theoretical analysis of sample complexity in symmetric composite binary quantum hypothesis testing for unknown quantum states.

Ax Mayank Sharma, Roy Pea, Hari Subramonyam 11d ago

ConvoLearn: A Learning Sciences Grounded Dataset for Fine-Tuning Dialogic AI Tutors

ConvoLearn: Dataset of 2,134 tutor-student dialogues for fine-tuning LLM-based AI tutors, grounded in dialogic learning theory and Earth Science curriculum.

Ax Bryan Sangwoo Kim, Jonghyun Park, Jong Chul Ye 11d ago

Tiled Prompts: Overcoming Prompt Misguidance in Image and Video Super-Resolution

Tiled Prompts: Method addressing prompt misguidance in text-conditioned diffusion models for image and video super-resolution by handling localized details.

Ax Cheng cheng, Chenxing Wang, Aolin Li, Haijun Wu, Huiyun Hu, Juyuan Wang 11d ago

When & How to Write for Personalized Demand-aware Query Rewriting in Video Search

WeWrite: Personalized query rewriting framework for video search systems using user history to identify search intent and resolve ambiguity.

Ax Daniel Zantedeschi, Kumar Muthuraman 11d ago

Fisher-Geometric Diffusion in Stochastic Gradient Descent: Optimal Rates, Oracle Complexity, and Information-Theoretic Limits

Theoretical analysis of stochastic gradient descent covariance under exchangeable mini-batch sampling and its connection to Fisher information.

Ax Yuanda Xu, Hejian Sang, Zhengze Zhou, Ran He, Zhipeng Wang 11d ago

PACED: Distillation and On-Policy Self-Distillation at the Frontier of Student Competence

PACED: LLM distillation method that weights training problems by student competence using gradient signal-to-noise ratio to improve distillation efficiency.

Ax Jiacheng Tang, Zhiyuan Zhou, Zhuolin He, Jia Zhang, Kai Zhang, Jian Pu 11d ago

CausalVAD: De-confounding End-to-End Autonomous Driving via Causal Intervention

Framework addressing causal confusion in end-to-end autonomous driving models through causal intervention during training to improve reliability and safety.

Ax Amirmohammad Farzaneh, Osvaldo Simeone 11d ago

Post-Selection Distributional Model Evaluation

Research on formal evaluation methods for machine learning models, focusing on test-time performance-reliability trade-offs when target KPI levels are unknown.

Ax Haochuan Kevin Wang, Zechen Zhang 11d ago

Kill-Chain Canaries: Stage-Level Tracking of Prompt Injection Across Attack Surfaces and Model Safety Tiers

Methodology for detecting prompt injection across multi-agent LLM pipelines. Stage-level kill-chain tracking for attack resilience evaluation.

Ax Yasaman Kashefbahrami, Erkut Akdag, Panagiotis Meletis, Evgeniya Balmashnova, Dip Goswami, Egor Bondarau 11d ago

R3PM-Net: Real-time, Robust, Real-world Point Matching Network

Point cloud registration network for 3D data. Deep learning approach for robust matching in real-world conditions.

Ax Reihaneh Zohrabi, Hosein Hasani, Akshita Gupta, Mahdieh Soleymani Baghshah, Anna Rohrbach, Marcus Rohrbach 11d ago

HaloProbe: Bayesian Detection and Mitigation of Object Hallucinations in Vision-Language Models

Detection and mitigation of object hallucinations in vision-language models. Bayesian approach analyzing attention weights and token confounders.

Ax Swarnadip Chatterjee, Vladimir Basic, Arrigo Capitanio, Orcun Goksel, Joakim Lindblad 11d ago

Needle in a Haystack: One-Class Representation Learning for Detecting Rare Malignant Cells in Computational Cytology

One-class learning for detecting rare malignant cells in medical images. Addresses class imbalance and limited annotations in cytology.

Ax Tao Han, Zhibin Wen, Zhenghao Chen, Fenghua Lin, Junyu Gao, Song Guo, Lei Bai 11d ago

Generative 3D Gaussian Splatting for Arbitrary-ResolutionAtmospheric Downscaling and Forecasting

3D Gaussian splatting for weather prediction downscaling. Proposes scale-aware vision transformer for arbitrary-resolution atmospheric forecasting.

Ax Seungjae Moon, Seunghyun Oh, Youngmin Ro 11d ago

OV-Stitcher: A Global Context-Aware Framework for Training-Free Open-Vocabulary Semantic Segmentation

Training-free semantic segmentation using vision-language models. Global context-aware framework for dense prediction without additional training.

Ax Nishikanta Mohanty, Bikash K. Behera, Badshah Mukherjee, Pravat Dash 11d ago

QARIMA: A Quantum Approach To Classical Time Series Analysis

Quantum-inspired ARIMA methodology for time series analysis. Combines quantum autocorrelation with variational circuits.

HN ingve 11d ago

Adventures in Slop: Can an AI Agent Generate Web Traffic?

Experiment using Claude to autonomously build a website designed to generate traffic, exploring AI agent capabilities and decision-making in open-ended tasks.

HN jeeybee 11d ago

Show HN: Rekal – Long-term memory for LLMs in a single SQLite file

MCP server enabling long-term memory for LLMs using SQLite, hybrid search (BM25+vectors), and local embeddings without API keys.

HN nradov 11d ago

Over 4,732 Messages, He Fell in Love with an AI Chatbot. Now He's Dead

Narrative article about user developing emotional attachment to AI chatbot.

HN robinw_ 11d ago

TokenMonopoly – Claude Max Is Overpriced; Compare AI Subscriptions

Live leaderboard comparing AI model subscriptions and API pricing across 27 benchmarked models from Claude, GPT, Gemini, DeepSeek, and others.

HN Input-X 11d ago

Been building a multi-agent framework in public for 5 weeks, its been a Journey

Multi-agent framework with persistent memory across sessions where agents collaborate on shared codebases and retain conversation context.

HN ingve 11d ago

Fun with an indecisive AI coding agent

Case study documenting indecisiveness in AI coding agent using Claude Opus 4.6 when debugging non-trivial bugs in GoAWK.

HN bgwmj 11d ago

Show HN: walnut – Error tracking AI agents

Error tracking tool designed specifically for AI agents with CLI interface, compatible with Sentry SDK for existing setups.

HN isitdan 11d ago

I built autonomous AI with memory and sleep, and it had nightmares

30-day experiment running autonomous AI system with memory and sleep cycles, documenting emergent behaviors and their implications.

HN ladino 11d ago

Show HN: Redactify – macOS/iOS app to redact sensitive data before using LLMs

macOS/iOS app automatically redacting sensitive personal, financial data, faces, and metadata before sharing documents with Claude and ChatGPT.

HN developerfred 11d ago

Nouns Agentic – Enabling AI Agents to Buy Nouns and Participate

Blockchain project enabling AI agents to participate in Nouns DAO governance.

HN s_brady 11d ago

Springdrift: An Auditable Persistent Runtime for LLM Agents

arXiv paper on Springdrift framework providing auditable persistent runtime environment for LLM agents.

HN cowartc 11d ago

The Three Enterprise Layers Are Collapsing into One

Enterprise architecture analysis on three-layer collapse in business process automation systems, discussing MCP servers and small LLM deployment.

HN __natty__ 11d ago

Worms 2 remastered 4k videos with no AI or upscaling

Announcement of Worms 2 remastered video collection without AI upscaling.

HN cyb_ 11d ago

What Claude Code's Source Revealed About AI Engineering Culture

Analysis of exposed Claude Code source revealing engineering practices: 259 PRs, 497 commits, 40K lines in 30 days, examining AI-assisted development culture.

HN obilgic 11d ago

Show HN: Posse – A Web UI for Claude Managed Agents

Posse is a web UI for Anthropic's Managed Agents, providing browser-based interface for agent creation, sessions, and memory management.

HN luu 11d ago

Tracking down a 25% Regression on LLVM RISC-V

Technical analysis of a 25% performance regression in LLVM RISC-V compiler optimization and fix implementation.

HN usestork 11d ago

Show HN: Stork – MCP server so Claude/Cursor can search 14k MCP servers AI tools

Stork.AI is a directory of 14k MCP servers and AI tools with community trust scores, offering a meta-MCP server for discovering integrations within Claude, Cursor, and other IDEs.

HN abbyedd 11d ago

Self-evolving daemon that 'dreams' about your codebase at night for AI agents

Entroly is a context compression engine that reduces LLM API costs by 80% for Claude, Cursor, and OpenAI by compressing codebase context without losing visibility.

HN mesto1 12d ago

Distributed AI Agents

Research on using distributed AI agents with independent context windows to improve reasoning on complex multi-perspective questions.

HN matisiekpl 12d ago

Show HN: NeonD – open-source DX-oriented Postgres Control Plane

NeonD is an open-source Postgres control plane based on NeonDB architecture with branching and PITR support.

HN FredrikMeyer 12d ago

AI Is Like TV

Opinion piece comparing AI adoption to TV, discussing shift to AI-assisted programming and loss of challenging side projects.

HN BlackVectorOps 12d ago

Standard Control Flow Flattening is fundamentally broken

Security research showing control flow flattening obfuscation in production SDKs is defeatable through static analysis.

HN Olshansky 12d ago

Core views on AI safety (March 2023)

Anthropic's 2023 statement on AI safety risks and impact, discussing concerns about powerful AI development in coming decade.

HN benswerd 12d ago

X for AI Agents

Minimal stub post with no content.

HN davidcann 12d ago

Show HN: Buildermark – See how much code is by your agents (open source, local)

Buildermark open-source tool measures code generation by AI agents vs human developers by matching agent conversations to git commits.

HN ninjagoo 12d ago

The race to train AI robots

Article on training AI robots to understand physical movement through human trainers in India and global industrial settings.

HN AISupportTeam 12d ago

I built a tool that checks if AI recommends your brand when buyers ask questions

Tool testing brand visibility across ChatGPT, Gemini, Claude, and other LLMs when users ask buying questions, with competitive analysis.

HN grimm8000 12d ago

Show HN: The AI agent that can improve itself

GrimmBot: Autonomous AI agent in sandboxed Docker with desktop/browser control, self-improvement capability, and tool building.

HN montyanderson 12d ago

1-bit inference of 0.8M param GPT running inside 8192 bytes of sram

1-bit quantized GPT with 800K parameters runs inference in 8KB of SRAM, demonstrating extreme model compression.

HN benmora 12d ago

Show HN:Lumisift – improves data retention in RAG from ~40% to 87%

Lumisift: Open-source tool improving data retention in RAG pipelines from 40% to 87% by fixing retrieval accuracy for scientific documents.