Reinforced Linear Genetic Programming

TL;DR

RLGP addresses learning register-action mappings by integrating Q-Learning with Linear Genetic Programming, yielding a hybrid approach that aims to improve search and adaptability. The authors present a memory-safe Rust framework with modular engines, integrate OpenAI Gym environments, and optimize both baseline LGP and RLGP using Optuna, evaluating on $cartpole-v1$ and $mountain-car-v0$ benchmarks. The results show that LGP can outperform RLGP on CartPole while RLGP struggles on MountainCar, underscoring the sensitivity to exploration strategies and problem structure. This work demonstrates a practical pathway for hybridizing evolutionary computation with reinforcement learning and provides a reusable platform for future EAs research.

Abstract

Linear Genetic Programming (LGP) is a powerful technique that allows for a variety of problems to be solved using a linear representation of programs. However, there still exists some limitations to the technique, such as the need for humans to explicitly map registers to actions. This thesis proposes a novel approach that uses Q-Learning on top of LGP, Reinforced Linear Genetic Programming (RLGP) to learn the optimal register-action assignments. In doing so, we introduce a new framework "linear-gp" written in memory-safe Rust that allows for extensive experimentation for future works.

Figures (4)

• Figure 1: Performance comparison of different genetic programming methods on iris dataset
• Figure 2: (Continued) Performance comparison of different genetic programming methods on iris dataset
• Figure 3: Comparison of RLGP vs LGP Performance on Cart Pole
• Figure 4: Comparison of RLGP vs LGP performance on Mountain Car