Tree-Structured Synergy of Large Language Models and Bayesian Optimization for Efficient CASH
LB-MCTS is a framework that automates algorithm selection and hyperparameter tuning (CASH problem) by using a Monte Carlo Tree Search tree as a shared state. It integrates Bayesian Optimization for quantitative searching and Large Language Models for semantic proposals. This synergy allows the system to dynamically shift between LLM and BO guidance, leading to superior performance on complex machine learning tasks while maintaining theoretical convergence guarantees.