Don't Play Games, Optimize [President's Message]
Yaochu Jin · IEEE Computational Intelligence Magazine · 2024
When I give a talk about evolutionary machine learning, one question I often expect is why I use an evolutionary algorithm to optimize the hyperparameters and structure of a neural network, rather than using a reinforcement learning algorithm. A quick answer might be, well, I am an evolutionary computation guy. I know this is a sloppy answer. Often, I attempt to explain the potential benefits of using an evolutionary algorithm in comparison with a reinforcement learning algorithm, e.g., in handling multiple objectives, in parallelizing the calculations, and also in dealing with sparse environmental feedback, among others. Clearly, it is always problem-dependent whether an evolutionary algorithm or a reinforcement learning algorithm should be chosen to solve a machine learning problem.