Reinforcement-Learning-and-DeepRL

There are two implementations of reinforcement learning.
One is classic RL including concepts and algorithms:
1. n arms bandit probelm - exploration and exploitation test with epsilon greedy policy
2. Value iteration v.s. policy evaluation and iteration.(based on dp, model-based)
3. Q-learning and Sarsa (model-free algorithms)
Another part is Deep RL where we implment DQN
DQN here is going to solve a famous Cartpole game. And we use 'gym' to get the game environment.
This part is based on tensorflow so two networks are built in the algorithm.
The script is run on Google-Colab platform.