Reinforcement Learning

Reinforcement Learning is another machine learning method that is based on the idea of learning by interacting with the environment. Reinforcement learning problems involve what to do or how to map situations to actions so as to maximize a reward signal in long term. A reinforcement learning problem is different from other machine learning methods that are based on learning from labeled data, A reinforcement learning agent learns from interaction with its encironment an the reward signal that it gets from the environment, not from labeled data, and from this prespective it is more similar to optimal control problems. Reinforcement learning research is certainly one of the most important part of the way toward the future of AI and HRI.

Fig1: The agent-environment interaction in reinforcement learning 1

Deep Reinforcement Learning

Link to the first few application of deep learning for classification of images are available here

Deep Q Network

The DQN (Deep Q-Network) algorithm was developed by DeepMind in 2015. DQN combining reinforcement learning and deep neural networks at scale. The algorithm was developed by enhancing a classic RL algorithm called Q-Learning with deep neural networks and a technique called experience replay.
Some of the resources available to learn Deep Q Network in concurrence with RL is here Understanding how DQN networks are made can be confusing and tiresome, one of the tutorials that takes time explaining all the layers for defining a deep network can be found here The tutorial is a five part series for understanding Reinforcement Learning with Python and extensively uses TensorFlow with Keras. With the new version of the Tensorflow, the available code may have some issue with the ModifiedTensorBoard. A functioning code in the Blob Environment is available [here] (https://github.com/sunamatya/Deep_learning_tutorials/blob/main/dqn-1.py)

## General Resources

• Sutton, Richard S., and Andrew G. Barto. Reinforcement learning: An introduction. MIT press, 2018

• Bertsekas, Dimitri. Reinforcement learning and optimal control. Athena Scientific, 2019

• Coursera: Reinforcement Learning Specialization, University of Alberta

• Deep learning with Keras and Tensor Flow

• Stefan Otte: Deep Neural Networks with PyTorch,PyData Berlin 2018

• Best tools and packages for reinforcement learning and deep reinforcement learning in python

• REINFORCEMENT LEARNING (DQN) with Pytorch, TUTORIAL

• REINFORCEMENT LEARNING (DQN) with TensorFlow, TUTORIAL
• Deep Q Learning and Deep Q Networks (DQN) Intro and Agent - Reinforcement Learning w/ Python Tutorial

Return Home Page

Main Topic 1

(EXAMPLE: Game Theory)
(Insert short introduction for main topic)

Concept or Subtopic 1

(EXAMPLE: Nash equilibrium solutions)
(ADD: description of concept or subtopic)

Sample Projects

(ADD: RISE Code/Github link and description)
(ADD: or in text example/walkthrough)

Additional Reading

• (ADD: Article)
• (ADD: Article)
• (ADD: Article)

Concept or Subtopic 2

(EXAMPLE: Pareto optimality)
(ADD: description of concept or subtopic)

Sample Projects

(ADD: RISE Code/Github link and description)
(ADD: or in text example/walkthrough)

Additional Reading

• (ADD: Article or resource)
• (ADD: Article or resource)
• (ADD: Article or resource)

Return Home Page

1. Image from: Sutton, Richard S., and Andrew G. Barto. Reinforcement learning: An introduction. MIT press, 2018 ↩