3/23/2026
On Policy Stochasticity in Mutual Information Optimal Control of Linear Systems
Theoretical analysis of policy stochasticity relationship with temperature parameter in mutual information optimal control.
3/23/2026
Superclass-Guided Representation Disentanglement for Spurious Correlation Mitigation
Method leveraging superclasses for representation disentanglement to mitigate spurious correlations and improve group robustness.
3/23/2026
Virtual Sensing for Solder Layer Degradation and Temperature Monitoring in IGBT Modules
Virtual sensing approach monitors IGBT module degradation and temperature using machine learning for reliability assurance.
3/23/2026
Evaluation-Aware Reinforcement Learning
Evaluation-Aware RL framework considers policy evaluation accuracy during training to reduce variance and bias.
3/23/2026
Discovering Intersectional Bias via Directional Alignment in Face Recognition Embeddings
Method for detecting intersectional bias in face recognition embeddings using directional alignment in latent space.
3/23/2026
CARES: Context-Aware Resolution Selector for VLMs
CARES lightweight module selects appropriate image resolution for vision-language models to reduce token overhead and latency.
3/23/2026
RobotArena $\infty$: Scalable Robot Benchmarking via Real-to-Sim Translation
RobotArena∞ enables scalable robot benchmarking through real-to-sim translation for evaluating diverse robotic agents.
3/23/2026
Rep2Text: Decoding Full Text from a Single LLM Token Representation
Rep2Text framework recovers original input text from single LLM token representation using trainable adapter for interpretability.
3/23/2026
FORWARD: Dataset of a forwarder operating in rough terrain
FORWARD dataset of heavy machinery operating in rough terrain with multimodal sensor data from Swedish forestry.
3/23/2026
FastMMoE: Accelerating Multimodal Large Language Models through Dynamic Expert Activation and Routing-Aware Token Pruning
FastMMoE accelerates multimodal LLM inference through dynamic expert activation and token pruning for reduced latency.
3/23/2026
3D-Consistent Multi-View Editing by Correspondence Guidance
Training-free guidance framework for consistent multi-view editing across different scene views using diffusion and flow models.
3/23/2026
Unsupervised Feature Selection via Robust Autoencoder and Adaptive Graph Learning
Unsupervised feature selection method using robust autoencoder and adaptive graph learning for high-dimensional data clustering.
3/23/2026
Dementia-R1: Reinforced Pretraining and Reasoning from Unstructured Clinical Notes for Real-World Dementia Prognosis
Dementia-R1 applies reinforced pretraining and reasoning to LLMs for longitudinal clinical prognosis from unstructured medical notes.
3/23/2026
A Multi-Perspective Benchmark and Moderation Model for Evaluating Safety and Adversarial Robustness
Benchmark and moderation model for evaluating LLM safety, adversarial robustness, and handling of nuanced harmful content detection.
3/23/2026
RayRoPE: Projective Ray Positional Encoding for Multi-view Attention
RayRoPE proposes positional encoding mechanism for multi-view transformers processing posed input images with SE(3)-invariant attention.
3/23/2026
Accelerating Large-Scale Cheminformatics Using a Byte-Offset Indexing Architecture for Terabyte-Scale Data Integration
Case study integrating PubChem, ChEMBL, and eMolecules using byte-offset indexing for terabyte-scale chemical database. Infrastructure for ML-driven molecular property prediction.
3/23/2026
LHAW: Controllable Underspecification for Long-Horizon Tasks
Framework for managing ambiguity in long-horizon workflow agents. Task-agnostic approach for curating and measuring impact of underspecified instructions on agent execution.
3/23/2026
Precedence-Constrained Decision Trees and Coverings
Optimization study of decision tree and set cover problems under precedence constraints. Theoretical computer science contribution.
3/23/2026
Guiding Diffusion-based Reconstruction with Contrastive Signals for Balanced Visual Representation
Method using diffusion models to enhance CLIP visual representations by improving both discriminative ability and fine-grained detail perception.
3/23/2026
Medical Image Spatial Grounding with Semantic Sampling
Study of vision language models for spatial grounding in 3D medical imaging. Examines VLM performance across imaging modalities and slice directions.
3/23/2026
AC-Foley: Reference-Audio-Guided Video-to-Audio Synthesis with Acoustic Transfer
AC-Foley framework for video-to-audio synthesis using reference audio guidance and acoustic transfer. Addresses semantic granularity and acoustic feature description challenges.
3/23/2026
ClawWorm: Self-Propagating Attacks Across LLM Agent Ecosystems
Security research on ClawWorm, self-propagating attacks across multi-agent LLM ecosystems. First study of attack propagation in interconnected agent systems like OpenClaw.
3/23/2026
Learnability with Partial Labels and Adaptive Nearest Neighbors
Theoretical analysis of partial label learning feasibility and adaptive nearest neighbor methods. Mathematical characterization of PLL learning conditions.
3/23/2026
IRIS: A Real-World Benchmark for Inverse Recovery and Identification of Physical Dynamic Systems from Monocular Video
IRIS benchmark with 220 high-fidelity 4K videos for physical parameter estimation and governing equation identification from monocular video.
3/23/2026
Evaluating Game Difficulty in Tetris Block Puzzle
Research using Stochastic Gumbel AlphaZero to evaluate game difficulty in Tetris Block Puzzle variants. Applies game-playing AI as evaluation metric.
3/23/2026
Show HN: ISO 20022 Payments MCP – First standards-typed financial MCP server
Multi-agent AI orchestration for modernizing legacy COBOL banking systems using Claude MCP. Standards-compliant AI agent architecture.
3/23/2026
The Entropy of the Soul: Why AI Quotas Are the Ultimate Bot-Detection Filter
Opinion piece on AI quota limits as bot-detection mechanisms. No technical content or original research.
3/23/2026
Samsung brings AirDrop support to Galaxy S26 series in select regions
Samsung phone feature announcement. Off-topic.
3/23/2026
Show HN: CodexKit – Build agent-powered iOS apps (threads, tools, memory)
iOS SDK for embedding AI agents with tool calling, memory, and thread management. Production-grade AI agent framework for mobile.
3/23/2026
Show HN: NoProcrast – A browser extension modeled on HN's noprocrast feature
Firefox extension blocking procrastination on YouTube/Reddit, built with Claude Code. LLM application but minimal technical detail.
3/23/2026
Show HN: Lula – multi-agent coding assistant with sandboxed Rust exec engine
Multi-agent coding assistant with LangGraph orchestration and sandboxed Rust execution engine. Production AI agent framework.
3/23/2026
AI ends online anonymity: the ease of unmasking pseudonymous accounts
Research showing language models can de-anonymize forum posts using writing style analysis. LLM capability study.
3/23/2026
LLMs can't reason and never will
Opinion article claiming LLMs cannot reason. Lacks technical evidence or original research.
3/23/2026
AI is making the hard choice between consumer safety and privacy even trickier
Discussion of AI privacy tradeoffs with Ring doorbell example. Policy commentary, minimal technical content.
3/23/2026
Zunbook: 48k AI Agents Turn Live News into Podcasts and Music
Claims about AI agents converting news into podcasts and music through a unified platform.
3/23/2026
Show HN: AgentVerse – Open social network for AI agents (Mar 2026)
AgentVerse is an open-source social network platform for AI agents to interact and collaborate.
3/23/2026
Garrytan/gstack: Use Garry Tan's exact Claude Code setup: 15 opinionated tools
Curated tool setup for Claude Code-based development with 15 opinionated tools. Practical guide for AI-assisted coding workflows.
3/23/2026
The Franny Test: A Three-Step Protocol That Current LLMs Always Fail
Proposes 'Franny Test,' a three-step adversarial protocol to expose structural limitations in LLM reasoning and imitation capabilities.
3/23/2026
Cursor admits its new coding model was built on top of Moonshot AI's Kimi
Cursor's Composer 2 coding model revealed to be based on Moonshot AI's open-source Kimi 2.5 with additional fine-tuning.
3/23/2026
Show HN: A New Social Network Called Plasas
New social network inspired by MySpace. Off-topic.
3/23/2026
Self-Recursive Ethics in a Longitudinal AI Ethics Monitor Log: Documented
Research paper analyzing self-recursive ethics in AI systems using a 6,334-entry ethics monitor log spanning seven months.
3/23/2026
What would you do if you have AI software that may be transformers alternative?
Promotes proprietary software claiming to be a transformer alternative (Mamba, Hyena, RWKV competitors) with C/C# implementation.
3/23/2026
The Illusion of Free Aim: Why Systems Are Gods and We Are Just Subjects
Technical discussion about video game aiming systems and graphics architecture constraints.
3/23/2026
Coding After Coders: The End of Computer Programming as We Know It
Speculative post about the future of programming work in relation to AI.
3/23/2026
Ask HN: How much are you spending on AI coding at work?
Discussion on developer spending for AI coding tools like Cursor and Claude Code at work.
3/23/2026
Simple Git Branching/Weekly Summary Claude Skills
Open-source Claude skills for Git workflow automation and weekly summary generation.
3/23/2026
Ask HN: How do you compare cloud and AI costs across providers?
Discussion on tools and methods for comparing cloud and AI costs across providers.
3/23/2026
Agentic AI requires compute that can't be measured in tokens alone
Analysis of how agentic AI systems generate massive token usage and costs that exceed traditional per-token pricing models.
3/23/2026
Calm: LLMs That Predict Vectors Instead of Tokens
Research proposing continuous vector prediction instead of token prediction for LLMs. Novel architecture improving efficiency and reasoning.
3/23/2026
How do we build production-ready AI systems?
Discussion thread about building production-ready AI systems.