Description

Reinforcement Learning (RL) possesses immense potential and is doubtless one of the most dynamic and stimulating fields of research in Artificial Intelligence. RL is considered as a game-changer in Data Science, particularly after observing the winnings of AI agents AlphaGo Zero and OpenAI Five against top human champions. However, RL is not restricted to games.

The progress in Reinforcement Learning, especially during the last few years, has been sensational. RL is everywhere now, ranging from resource management to chemistry, from healthcare to finance, and from Recommender Systems to more advanced applications in stock prediction.

Since RL is goal-oriented learning, an understanding of RL is not only vital but also indispensable in all the fields of Data Science. This course will enable you to take your career to the next level, as it presents you with a clear explanation of the concepts and implementations of RL in Data Science.

The course ‘Reinforcement Learning, Theory and Practice’ provides you with an opportunity for innovative, independent learning. The course focuses on the practical applications of RL and includes a hands-on project. The course is:

· Easy to understand.

· Descriptive.

· Comprehensive.

· Practical with live coding.

· Rich with advanced and the most recently discovered RL models by the champions in this field.

This course is designed for beginners, although complex concepts are covered later.

As this course is a compilation of all the basics, it will inspire you to move forward and experience much more than what you have learned. You will be assigned homework/ tasks/ activities at the end of each module, which will assess / (further build) your learning based on the concepts and methods you have learned earlier on. Since the aim is to get you up and running with implementations, many of these activities will be coding based.

Data Science is unquestionably a rewarding career. You get to solve some of the most interesting problems, and you are rewarded with a handsome salary package. A core understanding of RL will empower you with more AI tools and ensure progressive career growth.

As we have already said, RL possesses immense potential. Don’t miss out on this opportunity to learn the advanced concepts and methodologies of RL at a highly competitive price. The tutorials are subdivided into 75+ short HD videos along with detailed code notebooks.

Teaching is our passion:

Our online tutorials have been created with the best possible expertise to help you in understanding the RL concepts clearly. We have taken great care to ensure the code base is up to date. We really want you to accomplish a strong basic understanding of RL before you move onward to the advanced version. The perks of this compelling course include high-quality video content, assessment questions, meaningful course material, course notes, and handouts. You can also approach our team whenever you have any queries.

Course Content:

This all-inclusive course consists of the following topics:

1. Introduction

a. Motivation

i. What is Reinforcement Learning?

ii. How is it different from other Machine Learning Frameworks?

iii. Real-world examples

iv. Exercises and Thoughts

b. Terminology of Reinforcement Learning

i. Agent

ii. Environment

iii. Action

iv. State

v. Transition

vi. Reward

vii. Policy

viii. Exercises and Thoughts

c. Example Grid World

i. Deterministic World

ii. Stochastic World

iii. Stationary World

iv. Non-Stationary World

v. Exercises and Thoughts

2. Markov Decision Process (MDP)

a. Prerequisites

i. Probability Theory Review

ii. Modeling Uncertainty of Environment

iii. Running Averages

iv. Simulation in Python

v. Exercises and Thoughts

b. Elements of an MDP

i. Input: State Space

ii. Input: Action Space

iii. Input: Environment Model

iv. Input: Reward function

v. Output: Policy

vi. Worked Examples

vii. Exercises and Thoughts

c. More on Rewards

i. Delayed Reward

ii. Reward Scaling

iii. Policy Changes with Reward Scaling: Worked Example

iv. Infinite Horizons and Stationarity

v. Walks or Sequences

vi. Value of a Walk

vii. Stationarity of Preferences

viii. Discounted Rewards

ix. Exercises and Thoughts

d. Solving an MDP

i. Bellman Optimization Criteria

ii. Model-Based Value Iterations

iii. Optimal Value Function

iv. Finding Optimal Policy

v. Model-Based Policy Iterations

vi. Action-Value Functions

vii. Relationship Between Value Functions and Action-Value Functions

viii. Policy Evaluation

ix. Learner Evaluation

x. Exercises and Thoughts

3. Model Free Learning

a. Value Approximation

i. Episodes

ii. Running-Averages Applications

iii. Incremental Learning

iv. Properties of Learning Rates

v. Simulation in Python

vi. Exercises and Thoughts

b. Temporal Difference (TD) Learning

i. What is Temporal Difference?

ii. TD (1) Update Rule

iii. Eligibility Traces

iv. TD (1) Learning Algorithm

v. Implementation in Python

vi. Limitations of TD (1)

vii. Exercises and Thoughts

c. Toward TD(λ)

i. Maximum Likelihood Estimate

ii. TD (0) Update Rule

iii. TD (λ)

iv. K-Step Look-a-head

v. Combinations of Different Step Look-a-heads

vi. Good Values of λ

vii. TD (λ) Algorithm

viii. Implementation in Python

ix. Exercises and Thoughts

d. Q-Learning

i. Q-functions

ii. Contraction Mapping

iii. Bellman Operators

iv. Why Value Iteration Works?

v. Q-Learning Algorithm

vi. Implementation in Python

vii. Exercises and Thoughts

e. Policy Iteration

i. Direct Policy Learning

ii. Value Estimation in Policy Iteration

iii. Why Policy Iteration Works

iv. Policy Iteration Algorithm

v. Implementation in Python

vi. Exercises and Thoughts

4. Project

a. Game in OpenAI GYM

5. What Next?

a. Game Theory

b. How to Model Infinite States and Actions?

c. Deep Reinforcement Learning

After completing this course successfully, you will be able to:

   Understand how RL techniques are applied to resolve real-world problems.
   Understand the methodology of RL with Data Science using interesting examples.
   Complete a project on the OpenAI Gym toolkit.

Who this course is for:

   People who want to get their data speak.
   People who want to learn RL with real applications in Data Science.
   Individuals who are passionate about numbers and programming.
   People who want to learn Reinforcement Learning along with its implementation in realistic projects.
   Beginners in the field of Data Science and Artificial Intelligence

Requirements

   No prior knowledge is needed. You will start from the basics and gradually build your knowledge in the subject.
   A willingness to learn and practice.
   Knowledge of Python will be a plus.

Last Updated 12/2020