All Articles
16990 articles total
SpaceX falls to $135 IPO price ahead of Starship launch
SpaceX stock price declines as market expectations cool following its IPO.
Windows 0-day drops the same day Microsoft releases record number of patches
A critical Windows 0-day vulnerability, identified as HiveLegacy, was discovered on the same day Microsoft released a massive patch update.
Diversity-Enriched Option-Critic
Introduction of a Diversity-Enriched Option-Critic framework that uses information-theoretic intrinsic rewards to improve behavioral diversity in reinforcement learning agents.
Seeing Through Uncertainty: Free-Energy-Inspired Real-Time Adaptation for Robust Visual Navigation
FEP-Nav is a biologically inspired framework that uses the Free Energy Principle to enable real-time perceptual adaptation for robust visual navigation in robots.
Enabling Energy-Efficient Simultaneous Multi-Task Reinforcement Learning through Spiking Neural Networks with Active Dendrites for Bio-inspired Generalist Agents
MTSpark is a novel methodology combining Spiking Neural Networks and Reinforcement Learning to create energy-efficient, multi-task generalist agents.
Modeling Story Expectations: A Generative Framework using LLMs
A generative framework using LLMs to model and predict consumer story expectations based on narrative content.
Voxatron
A discussion regarding Voxatron, though content is limited to comments.
Daniel Ek’s body-scanning startup Neko Health raises another $700M
Daniel Ek's Neko Health startup has raised $700M to develop proprietary body-scanning technology for health assessment.
Roblox is shutting down its video chat service
Roblox is discontinuing its Roblox Connect video chat service and its developer APIs due to low usage.
AI slop movies are the new direct-to-video cash grabs
Film studio Fountain 0 is utilizing AI to generate a reimagining of The Odyssey, reflecting a trend of AI-generated 'slop' movies.
Measurement Risk in Supervised Financial NLP: Rubric and Metric Sensitivity on JF-ICR
Researchers analyze measurement risk in supervised financial NLP, highlighting how rubric wording and metric choice significantly affect LLM evaluation.
Attractor Geometry of Transformer Memory: From Conflict Arbitration to Confident Hallucination
A study on the attractor geometry of Transformer memory reveals that 'conflict' and 'hallucination' share a unified geometric account in hidden-state space.
From Controlled to the Wild: Evaluation of Pentesting Agents for the Real-World
The paper introduces a practical evaluation protocol for AI pentesting agents to move beyond CTF benchmarks toward real-world vulnerability discovery.
Brain Vascular Age Prediction Using Cerebral Blood Flow Velocity and Machine Learning Algorithms
A study uses TCD and machine learning to predict brain vascular age, identifying accelerated aging in subjects with neurological diseases.
Matilda: Engine-Agnostic Search with Human Policy Guidance
Matilda is introduced as a set transformer that re-ranks legal-move distributions from Maia-3 to improve high-Elo chess play and individual style modeling.
Propheticus: Machine Learning Framework for the Development of Predictive Models for Reliable and Secure Software
Propheticus is presented as a machine learning framework designed to simplify the creation of predictive models for software reliability and security.
Duskers, the scary command line game, is getting a sequel
The command-line based horror game Duskers is receiving a sequel, continuing its unique gameplay mechanic of controlling ships via a terminal.
Show HN: Capn-hook for coding agents – don't grep the same mystery twice
Capn-hook is introduced as a tool for coding agents to avoid redundant grep operations by remembering previous search results.
Amid hardware legal battle, OpenAI releases a $230 keyboard for Codex
OpenAI has released a specialized light-up keyboard designed for its Codex agentic coding application amid an ongoing legal dispute with Apple.
Mobility-Aware Cache Framework for Scalable LLM-Based Human Mobility Simulation
The MobCache framework is proposed to enable efficient large-scale human mobility simulations by using reconstructible caches to reduce LLM computational costs.