Overview
Deep Reinforcement Learning refers to the ability of an “artificial agent” to learn by trial-and-error and rewards-and-punishments. An artificial agent aims to emulate a human’s ability to obtain and construct knowledge on its own, directly from raw inputs such as vision. To realize reinforcement learning, deep learning and neural networks are used. Reinforcement learning is different from machine learning and does not rely on supervised and unsupervised learning approaches.
In this instructor-led, live training, participants will learn the fundamentals of Deep Reinforcement Learning as they step through the creation of a Deep Learning Agent.
By the end of this training, participants will be able to:
- Understand the key concepts behind Deep Reinforcement Learning and be able to distinguish it from Machine Learning
- Apply advanced Reinforcement Learning algorithms to solve real-world problems
- Build a Deep Learning Agent
Audience
- Developers
- Data Scientists
Format of the course
- Part lecture, part discussion, exercises and heavy hands-on practice
Requirements
- Proficiency in Python
- An understanding of college Calculus and Linear Algebra
- Basic understanding of Probability and Statistics
- Experience creating machine learning models in Python and Numpy
Course Outline
Introduction
Reinforcement Learning Basics
Basic Reinforcement Learning Techniques
Introduction to BURLAP
Convergence of Value and Policy Iteration
Reward Shaping
Exploration
Generalization
Partially Observable MDPs
Options
Logistics
TD Lambda
Policy Gradients
Deep Q-Learning
Topics in Game Theory
Summary and Conclusion