/Analogy_of_Reinforcement_learning

Primary LanguageJupyter Notebook

Udacity Machine Learning Engineer Nanodegree Capstone Project

Requirements:

  1. python 3.5+
  2. openAi.gym (pip install gym)

Running the agents

use python <agent_name>.py to execute the agent. The list of various agents is given in the project structure\code of this Read_Me.

project Structure

Code

Cartpole

  1. Random_exploration : random_guessing.py : shows results if the env is enxplored randomly without any learning agent. (code\cartpole\random_guessing.py)
  2. Q_learning: Q_learning_agent.py : Creates apolicy iteration Q_learning model and genrates the respective Q_table while learning and produces the reeepctive results.(code\cartpole\Q_learning_agent.py)
  3. Deep_Q_learning : DQN_model_1.py : Creates a Deep_Q_learning agent and trains it to play the game. (code\cartpole\DQN_model_1.py)

Mountain_Car

  1. Random_exploration : random_guessing.py : shows results if the env is enxplored randomly without any learning agent.(code\Mountain_car\random_guessing.py)
  2. Q_learning: Q_learning_agent.py : Creates apolicy iteration Q_learning model and genrates the respective Q_table while learning and produces the reeepctive results.(code\Mountain_car\Q_learning_agent.py)
  3. Deep_Q_learning : DQN_model_1.py : Creates a Deep_Q_learning agent and trains it to play the game.(code\Mountain_car\DQN_model_2.py)

Report

  1. Proposal : Docs/proposal.pdf
  2. Report : Docs/Report.html

Simulation Results :

Cartpole :

  1. Jupyter Notebook : Docs/Cartpole Results/Cartpole_result_plots.ipynb
  2. HTML : Docs/Cartpole_result_plots.html

Mountain Car :

  1. Jupyter Notebook : Docs/Mountain Car Results/Mountain_car_result_plots.ipynb
  2. HTML : Docs/Mountain_car_result_plots.html