AutoMind: Adaptive Knowledgeable Agent for Automated Data Science
AUTOMIND is an adaptive LLM agent framework designed for automated data science by overcoming limitations of rigid workflows and inflexible coding. It introduces three key advances: an expert knowledge base for domain expertise, a tree search algorithm to explore solutions, and a self-adaptive coding strategy that tailors code generation to task complexity. Evaluations show it outperforms state-of-the-art baselines.