Fitness Approximation through Machine Learning
This work addresses the issue of costly fitness computation in evolutionary algorithms, particularly for domains like game simulators, but it is incremental as it builds on existing fitness approximation methods.
The paper tackles the problem of reducing runtime in genetic algorithms by using machine learning models to approximate fitness, achieving significant improvements in evolutionary runtimes with fitness scores that are either identical or slightly lower than those from fully computed runs, depending on the approximation ratio.
We present a novel approach to performing fitness approximation in genetic algorithms (GAs) using machine-learning (ML) models, through dynamic adaptation to the evolutionary state. Maintaining a dataset of sampled individuals along with their actual fitness scores, we continually update a fitness-approximation ML model throughout an evolutionary run. We compare different methods for: 1) switching between actual and approximate fitness, 2) sampling the population, and 3) weighting the samples. Experimental findings demonstrate significant improvement in evolutionary runtimes, with fitness scores that are either identical or slightly lower than that of the fully run GA -- depending on the ratio of approximate-to-actual-fitness computation. Although we focus on evolutionary agents in Gymnasium (game) simulators -- where fitness computation is costly -- our approach is generic and can be easily applied to many different domains.