All Articles
16066 articles total
FPEdit: Robust LLM Fingerprinting through Localized Parameter Editing
FPEdit is a framework for robust LLM fingerprinting using localized parameter editing to protect intellectual property.
Balancing Privacy and Efficiency: Music Information Retrieval via Additive Homomorphic Encryption
A study on using additive homomorphic encryption to balance privacy and efficiency in Music Information Retrieval systems.
A simple clustering algorithm for lists
A discussion on a simple clustering algorithm for lists, providing a lightweight alternative for organizing data sequences.
Dario Amodei's stance on open weights is self-serving and short-sighted
A critical perspective on Dario Amodei's views on open weights for AI models, arguing they are self-serving and short-sighted.
Lerd, an open source Herd-like PHP development environment for Linux and macOS
Lerd is an open-source PHP development environment for Linux and macOS, designed as an alternative to Herd.
Agent-UCT: Upper Confidence Bounds Applied to Trees for Agentic Workflow Optimization with Cost-Awareness
Introduction of Agent-UCT, a cost-aware tree search algorithm for optimizing agentic workflows like RAG pipelines, along with the RAGSpace and WTB frameworks.
TRACE-CTI: Auditable Post-Extraction Governance of TTP Claims with Knowledge Graphs
TRACE-CTI is a post-extraction governance framework for Cyber Threat Intelligence using knowledge graphs to ensure auditable TTP claims.
Pushing the Frontier on Approximate EFX Allocations
Research on approximate envy-freeness up to any good (EFX) allocations for indivisible goods among agents with additive valuation functions.
One-Frame Calibration with Siamese Network in Facial Action Unit Recognition
Proposes a Calibrating Siamese Network (CSN) for facial action unit recognition using one-frame calibration to mitigate identity bias.
The Reliability of LLMs for Medical Diagnosis: An Examination of Consistency, Manipulation, and Contextual Awareness
An examination of LLM reliability for medical diagnosis, finding that models are susceptible to irrelevant prompt manipulation and context shifts.
Task and Skill Planning: Hierarchical Robot Planning with Black-Box Skills
Introduces Task and Skill Planning (TASP) for hierarchical robot planning, integrating black-box policies and Composable Interaction Primitives (CIPs).
When Should AI Follow? Task Structure and Joint Adaptation by Human and AI Agents
A computational model examining human-AI joint sequential adaptation, suggesting that AI should follow high-performing humans rather than always being deployed first.
June in Servo: real world compat, media queries, SharedWorker, and more
Servo provides an update on their progress in June, highlighting improvements in real-world compatibility, media queries, and SharedWorker support.
Nuclear Waste Cleanup: DOE Is Missing Opportunities to Apply Lessons
A report suggests the Department of Energy is missing opportunities to apply lessons learned in nuclear waste cleanup.
Google nixes its Earth AI feature one day after launch, amid criticism it would spread misinformation
Google quickly removed an AI feature for Google Earth that allowed users to generate fake imagery due to misinformation concerns.
Fresh off its Wiz payout, Index Ventures raises $2B across three funds
Index Ventures has raised $2 billion across three new funds, bringing their total available capital to $3.5 billion.
Apple’s new AirTags are back down to their best price
Apple's second-generation AirTags are currently available at discounted prices through several major retailers.
How is your enterprise tracking AI agent telemetry? Groundcover thinks it should never leave your cloud
Observability startup groundcover has raised $100M to expand its eBPF-based, bring-your-own-cloud (BYOC) telemetry platform for AI agents.
The Capability Paradox: How Smarter Auditors Make Multi-Agent Systems Less Secure
Researchers identify a 'capability paradox' where stronger AI agents in multi-agent systems can actually increase the attack success rate of semantic hijacking attacks.
Ratchet: A Minimal Hygiene Recipe for Self-Evolving LLM Agents
The Ratchet framework introduces a hygiene recipe for self-evolving LLM agents to better manage the lifecycle of their natural-language skill libraries.