Fuzzy Aggregated Topology Evolution for Cognitive Multi-tasks

Evolutionary optimization aims to tune the hyper-parameters during learning in a computationally fast manner. For of multi-task problems, evolution is done by creating unified search space with dimensionality that can include all tasks. Multi-task achieved via selective imitation where two individuals same type skill are encouraged crossover. Due relatedness tasks, resulting offspring may have for different task. In this way, we simultaneously evolve population excel paper, consider called Genetic Programming (GP) genes tree-like structure and be lengths hence naturally represent multiple We apply model neuroevolution determine optimal neural network such as number nodes, rate, training epochs using evolution. Here each gene encoded hyper parameters single network. Previously, was enabling or disabling individual connections between neurons This method extremely slow does not generalize well new architectures Seq2Seq. To overcome limitation, follow modular approach sub-tree GP sub-neural architecture preserved crossover across Lastly, order leverage on inter-task covariance faster evolutionary search, project features from both tasks common fuzzy membership functions. The proposed used topology feed-forward classification emotions physiological heart signals also Seq2seq chatbot converse kindergarten children. outperform baselines over 10% accuracy.