Machine Learning Based Rubik's Cube Solvers This project aims to implement the following:
- A Rubik's cube object which can be interacted with. Additionally a Rubik's cube environment for RL based algorithms.
- A solver based on a Deep Neural Network (not RNN, as Rubik's cube is essentially memoryless)
- A Q-Network based solver with limited states A Generic implementation of a Q-Network object and a Q-Netowrk training algorithm which takes in any environment with finite states
- A DQN based full cube solver A Generic implementation of an DQN object and DQN training algorithm which takes in any environment with finite states