Memory-Guided Trust-Region Bayesian Optimization (MG-TuRBO) for High Dimensions

Build a robust library for high-dimensional Bayesian optimization that incorporates memory-based trust regions. This tool would be highly valuable for calibrating simulation-based digital twins in robotics and logistics.

Suggested repo: turbo-mem

"Optimize complex, noisy simulations with memory-efficient Bayesian search."

Estimated effort: 80h