Mastering Unity Machine Learning Agents (Application Edition)

Lecture Topics 📖

In this Unity Machine Learning Agent Complete Mastery (Applied) lecture, you will learn the following!

• How to create a reinforcement learning environment using Unity
• How to apply machine learning agents for applied reinforcement learning techniques
• Applied Reinforcement Learning Algorithm Theory and Code Writing Method
• How to learn applied reinforcement learning algorithms using mlagents-learn

Lecture Features ✨

This lecture covers not only the theory of reinforcement learning and writing code, but also how to create a reinforcement learning environment to learn the corresponding algorithm, so it can be said to cover all the contents for reinforcement learning.
In addition, this course covers a wide range of reinforcement learning applications, including general reinforcement learning algorithms, multi-agent, curriculum learning, distributed learning, and difficult exploration environments.

What you'll learn 📚

In the Complete Guide to Unity Machine Learning Agents (Applied Edition), you will learn about the creation of environments for applied reinforcement learning techniques, reinforcement learning algorithm theory, and code content.
Specifically, the topics covered in this lecture are as follows:

Algorithm

• Proximal Policy Optimization (PPO)
• Attention PPO
• Adversarial PPO
• MA-POCA
• Exploration by RND (Random Network Distillation)
• HyperNetworks

environment

• Dodge
• Pong
• EscapeRoom
• Maze
• TwoMission

The videos below are the reinforcement learning environments you will implement yourself in this lecture and the results of learning through the reinforcement learning algorithms you will implement.

Things to note before taking the class 📢

The content of this lecture is explained based on the assumption that you have completed Inflearn's " Unity Machine Learning Agent Complete Mastery (Basic Edition)" ! ( Basic Edition Link )
If you haven't taken the basic course, I recommend taking the basic course first! However, if you know the basics of reinforcement learning, Unity, or how to use machine learning agents, you can take the applied course right away.

Practice environment

In the case of the practice environment, since the version in which the lecture was conducted is a bit older, we checked compatibility with relatively new software! You can proceed with the lecture content with either the "Lecture Progress Version" or the "Compatibility Check Version" setting below.

Practice environment (lecture version)

• Windows 10
• Unity 2021.1.18
• ML-Agents 2.0 (Unity) / ML-Agents 0.26.0 (Python)
• Python 3.8
• Pytorch 2.0

Practice environment (compatibility check version)

• Windows 10
• Unity 2022.3.4
• ML-Agents 3.0 (Unity) / ML-Agents 1.0.0 (Python)
• Python 3.8
• Pytorch 2.1

GitHub 🐙

You can check out all the environments and algorithm codes for this lecture on the following GitHub! Frequently asked questions are also organized on the GitHub wiki, so please refer to them!
https://github.com/reinforcement-learning-kr/Unity_ML_Agents_2.0

Course Target Audience/Course Purpose 🙆‍♀️

Types of students that knowledge sharers think of

• Developers interested in developing reinforcement learning environments
• Students and researchers interested in the theory and implementation of reinforcement learning.