Research Summer Camp (RSC) is a set of research projects made within the School of AI Algiers club SOAI LinkedIn that included different research topics, this notebook includes the implementation been done on Automatic Curriculum Learning (AutoCL) for Hard Exploration Environments
AutoCL is a technique used within Reinforcement Learning applications especially while tackling hard exploration environments that raise certain difficulties for RL algorithms to learn from scratch.
The environment that has been chosen to conduct our research and implementation is Rubik's cube. The latter offers the challenge of solving a sparse reward problem, which means that the agent will only receive a reward of 1 when the cube is solved and 0 otherwise. This makes the learning process very hard and time-consuming. We used the Rubik's Cube Jumanji Environment