Neural Networks based Hand-crafted genetic learning approach to simulate Space Mario Game

摘要

Genetic algorithm is a search heuristic inspired by Charles Darwin’s theory of natural selection where the fittest individuals pass on their traits to the next generation. The notion behind Natural Selection is to select the fittest individuals from a population-based on a fitness score and crossing over their features to create better off-springs and adding some mutation. The process repeats itself until the desired generation score is reached. To simulate a space Mario game, our research proposes a handcrafted genetic learning model that selects the best surviving entities from a population, crossing them over according to fitness priorities. The fittest entity from the previous generation is also passed along with some mutation. This continues until the fittest entities reach the target score in fewer generations possible. Our model crosses the target score of 2000 in the 16th generation.