Machine Learning Course Project

Jan 12, 2026ยท
Yang Zhang
,
Yanru Guan
,
Yuetian Wu
,
Zhi Zhang
ยท 1 min read

Overview

This project explores the use of large language model (LLM)-based program evolution to optimize geometric extremal problems, specifically minimizing the ratio of maximum to minimum pairwise distances in a set of points. It introduces an Experience Replay Memory (ERM) mechanism to improve the OpenEvolve framework, significantly accelerating convergence and improving solution quality through human-LLM collaboration.

Report

Blogs