I am working on a project using NEAT (NeuroEvolution) for malware classification. The dataset is relatively large - say 70000 training and 50000 test samples, and I imagine that evaluating the fitness function, which would be a cost function for each genome after passing all samples through the network would take a long time. Is there any way I could speed up the evaluation of the fitness of each genome or use a subset of samples in the training data. If so, how would the sampling process work? Any guidance greatly appreciated Continue reading...
