AI/ML arXiv cs.AI

Expert Behavior Prior Reinforcement Learning

Proposed Expert Behavior Prior (EBP) algorithm to improve online reinforcement learning sample efficiency using a Q-guided conditional variational autoencoder.

Other arXiv cs.AI

Regulating autonomous and agentic AI

A study on the challenges of regulating autonomous and agentic AI, proposing a shift from reactive to active regulatory systems.

AI/ML arXiv cs.AI

SPORD: A Simulation-Propose-then-OR-Dispose Approach for Supply Chain Planning

The SPORD method and NetSim platform by JD.com for decoupled supply chain planning using simulation and integer programming.

AI/ML arXiv cs.AI

Towards Faithful Graph Explanations with Synergistic Edge Effects via Granular Balls

Introduction of SeeExplainer, a parameter-free explainer for Graph Neural Networks that captures synergistic edge effects via granular balls.

AI/ML arXiv cs.AI

Multimodal Pretraining for Generalizable EEG Representation Learning

A multimodal EEG foundation model utilizing Mamba architecture and Vision Transformers for generalizable seizure detection.

AI/ML arXiv cs.AI

MSBraM: A Multi-scale Self-supervised Brain Foundation Model for Hierarchical EEG Dynamics Learning

MSBraM, a multi-scale self-supervised brain foundation model designed to learn hierarchical EEG representations through a two-stage pretraining framework.

AI/ML Hacker News

Flux 3 X Mimic: The Next Generation of Video-Action Models

Discusses Flux 3 X Mimic, a new generation of video-action models.

Other Hacker News

Nothing works and everyone is euphoric

A reflective or critical piece on the current state of technology or society where 'nothing works' despite general euphoria.

Other Hacker News

It's getting harder to focus every day

An exploration or discussion regarding the increasing difficulty of maintaining focus in the modern era.

Software Engineering Hacker News

Show HN: A static, client-side tarot calculator for Brazilian Portuguese

A showcase of a static, client-side tarot calculator specifically for Brazilian Portuguese users.

Other Hacker News

Future euro banknote design proposals

News regarding design proposals for future euro banknotes.

Other Hacker News

The Book Prize Index

An article or index focusing on book prizes.

Software Engineering Hacker News

I could do that in a weekend (2016)

A retrospective piece from 2016 discussing the common fallacy of thinking a complex project can be done 'in a weekend'.

AI/ML arXiv cs.AI

A New Well-Supported Semantics for Description Logic Programs

Proposes a new well-supported semantics for Description Logic Programs to reduce computational complexity of the consistency problem to NP-complete.

AI/ML arXiv cs.AI

How Rules Represent Causal Knowledge: Causal Modeling with Probabilistic Logic Programming

Integrates Pearl's causal modeling theory into probabilistic logic programming to enable the prediction of intervention effects in non-acyclic relationships.

AI/ML arXiv cs.AI

ICAE-Bench: Evaluating Coding Agents as Interactive Project Builders

Introduces ICAE-Bench, a new benchmark for evaluating coding agents on their ability to build projects from fuzzy, interactive requirements.

Other Hacker News

The day Steve Jobs dissed me in a keynote (2010)

A personal account of an encounter with Steve Jobs during a 2010 keynote presentation.

Cybersecurity The Verge

Microsoft pressures LG into killing unwanted McAfee ads

Microsoft successfully pressured LG to disable intrusive McAfee antivirus trial pop-ups that were appearing on Windows 11 PCs connected to LG monitors.

AI/ML arXiv cs.AI

AttriMem: Attribution-Guided Process Feedback for Agent Memory Learning

Researchers propose AttriMem, an RL-based framework that uses token-level attribution to provide fine-grained process feedback for improving LLM agent memory construction.

AI/ML arXiv cs.AI

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure

V-DEAL is a diagnostic framework designed to identify and mitigate safety de-calibration in Video LLMs, specifically focusing on how harmful videos can bypass safety filters when paired with benign queries.