Go-Explore is a reinforcement learning algorithm designed to find optimal solutions to complex problems that have a large action and state space. The algorithm was developed by Adrien Ecoffet and uses evolutionary algorithms and machine learning techniques to efficiently find optimal solutions. It begins by exploring a large number of random paths and then uses an evolutionary algorithm to store the best solutions found and combine them to create new paths. Go-Explore has the advantage of being able to find optimal solutions in complex and intractable problems where other reinforcement learning algorithms might fail.
