AutoReproduce: AI-Powered Automatic Experiment Reproduction
AutoReproduce automates AI experiment reproduction using paper lineage and multi-agent frameworks, enhancing research reliability and efficiency.
Local Markov Equivalence for Efficient Causal Discovery
Discover controlled direct effects efficiently using Local Markov Equivalence and PC-style algorithms with reduced tests and weaker assumptions.
Do LLMs Struggle with Math in Different Cultures?
Study reveals LLMs' math accuracy drops in culturally diverse contexts, urging diverse data for better global performance.
Survey of Computer Use Agents: Challenges & Future Trends
Explore key challenges and future directions in Agents for Computer Use (ACUs) to improve automation, learning, and real-world application.
Tractability Frontier for Exact Relevance Certification in AI
Explore the tractability frontier for exact relevance certification in AI decision-making, revealing key findings and meta-impossibility theorems.