April Edition: Reinforcement Learning

              1. OpenAI Gym from scratch

              2. Reinforcement Learning with Python

              3. Training Bots to Play Tennis

              4. Learning to Drive Smoothly in Minutes

              5. Robotic Control with Graph Networks

          Reinforcement learning (RL) refers to goal-oriented algorithms, in which an “agent” learns to accomplish a specific objective or goal while maximizing a set “reward” over many steps. RL agents are what people often think of when they describe “Artificial Intelligence” as portrayed in movies. The RL agent often starts from a blank slate, and under the right conditions, can achieve amazing performances as it learns about its environment. The RL algorithms are penalized when they make the wrong decisions and rewarded when they make the right ones, hence the term “reinforcement learning”. Setting up the environment, choosing the appropriate algorithm/strategy, and designing a reward function that will provide the incentives to achieve the desired behavior from the agent — these are all some of the aspects that make Reinforcement Learning both a fascinating and complex field.

Applications of Reinforcement Learning are varied across multiple industries including Robotics, Chatbots, and Self-Driving Cars. Most new students of Reinforcement Learning get introduced to RL by using the OpenAI Gym framework. TDS authors, Jonathas Figueiredo and Vihar Kurama wrote excellent articles on OpenAI and RL using Python to get you started. Once you have grounded your knowledge in RL, then have a look at more complex applications such as Training Agents to Play Tennis or Training a Chatbot using RL.

          While Convolution Neural Network (CNN) and Recurrent Neural Network (RNN) are becoming more important for businesses due to their applications in Computer Vision (CV) and Natural Language Processing (NLP), Reinforcement Learning (RL) as a framework for computational neuroscience to model decision making process seems to be undervalued.

          There is a lot of work and tutorials out there explaining how to use the OpenAI Gym toolkit and also how to use Keras and TensorFlow to train existing environments using some existing OpenAI Gym structures. However, in this tutorial, I will explain how to create an OpenAI environment from scratch and train an agent on it.

          Reinforcement is a class of machine learning where an agent learns how to behave in the environment by performing actions and thereby drawing intuitions and seeing the results. In this article, you’ll learn to understand and design a reinforcement learning problem and solve it in Python.

          This post explores my work on the final project for Udacity’s Deep Reinforcement Learning Nanodegree. My goal is to help other machine learning (ML) students and professionals who are in the early phases of building their intuition in reinforcement learning (RL).

        In this post, we will see how to train an autonomous racing car in minutes and how to smooth its control. The method, based on Reinforcement Learning (RL) and presented here in simulation (Donkey Car simulator), was designed to be applicable in the real world. It builds on the work of a startup named Wayve.ai that focuses on autonomous driving.

           Machine learning is helping to transform many fields across diverse industries, as anyone interested in technology undoubtedly knows. Things like computer vision and natural language processing were changed dramatically due to deep learning algorithms in the past few years, and the effects of that change are seeping into our daily lives.

  

Enjoyed reading this blog? Then why not share it with others. Help us make this AI community stronger. 

To learn more about such concepts related to Artificial Intelligence, visit our insideAIML blog page.

You can also ask direct queries related to Artificial Intelligence, Deep Learning, Data Science and Machine Learning on our live insideAIML discussion forum.

Keep Learning. Keep Growing.