KerasGenetic allows you to leverage the elegent modeling API Keras while performing training with genetic algorithms. Typically, Keras neural network weights are optimized by minimizing a loss function through the process of gradient descent.
Keras Genetic takes a different approach to weight optimization by leveraging genetic algorithms. Genetic algorithms allow you to optimize a neural network without in scenarios where there is no information about the loss landscape.
Genetic algorithms can be used to train neural networks in niche cases where you need to train a specialized controller with <1000 parameters.