Used Keras or PyTorch? These frameworks make it so easy to build and train Deep Neural Networks.
New Deep RL frameworks like Ray RLlib have made it equally easy to build and train Deep RL agents. Prototyping Deep RL agents now takes hours instead of days.
In this course, you will progress from no knowledge of RL to building powerful Deep RL agents using Ray RLlib, in just 4 hours.
For example, look at this robot, which is so dumb that it can't even stand on its two legs.
After just one evening of lessons and exercises, you will be able to teach this robot to walk using Deep Reinforcement Learning.
In the process, you will also learn
- Core concepts of RL
- How to identify problems in your industry and life where you can apply RL
- How to use industry leading Deep RL tools Ray RLlib and OpenAI Gym to solve simple RL tasks
You must have seen the news about superhuman Deep RL gaming agents like Atari DQN, AlphaGo, and OpenAI Five. You probably also heard about fascinating commercial successes of Deep RL like Data Center Cooling, Financial Trading, and Warehouse Automation.
Building these powerful Deep RL agents with your own hands is an exhilarating experience. I built this course so that you can experience that joy in the shortest time possible.
Plus, you'll develop solid conceptual foundation in RL, applicable practical skills, and a sharp eye for application opportunities in the industry.
I am looking forward to seeing you inside. Let's have some fun building Deep RL agents!
*free until 08.07.2022
"This course broke down complex RL concepts into small pieces that I could easily understand"
Martin Musiol - Managing Data Scientist, IBM
"Brilliant introduction to RL concepts and how they map to RLlib."
Jules S. Damji - Lead Developer Advocate, Anyscale
Who is this course for?
- Primarily for Data Scientists and Data Engineers, who want to use Deep RL
- Suitable for managers who want to know how Deep RL is applied in the industry and have an overview of standard Deep RL tools
- Suitable for students in a university-level Machine Learning curriculum, who want a practical introduction to Deep RL
Prerequisites
- You must have a background in Machine Learning to benefit from this course. In particular, you should already know the basic concepts of Machine Learning and Deep Learning. Ideally, you have used frameworks like Keras or PyTorch before.
- You are able to write object oriented code in Python and have basic knowledge of NumPy.
What will you learn?
Chapter 1
In Chapter 1, you will get introduced to RL and understand its key concepts.
- You will be able to decide if RL or supervised/self-supervised methods is the better tool for a given learning task
- You will analyze case studies of successful RL applications in the industry
Chapter 2
In Chapter 2, you will learn the steps and tools involved in solving Reinforcement Learning tasks.
Chapter 3
In Chapter 3, we will focus on RL environments. You will get introduced to OpenAI Gym and its API. By the end of this chapter, you will be able to control agents in Gym environments.
Chapter 4
In Chapter 4, we will focus on Deep RL frameworks.
- You will use Ray RLlib's implementation of Deep RL algorithms to solve OpenAI Gym environments
- You will know how Deep RL algorithms work at the high level
- You will learn to choose algorithms (e.g. DQN, PPO, TD3, SAC etc.) appropriately based on the nature of the problem
- You will be able to visualize training and evaluation metrics using Tensorboard
- You will be able to save trained agents and use them later (and even capture videos of live agents)
By the end of this chapter, you will be able to solve most benchmark RL tasks in OpenAI Gym.
Detailed Curriculum*
*Some lessons can be previewed without enrolling
- What is Reinforcement Learning? (5:46)
- Quiz: RL Objective
- Visualizing Reinforcement Learning Tasks (5:47)
- Reinforcement Learning vs. Supervised/Self-Supervised Learning (8:49)
- Quiz: Problem Attributes That Justify RL Application
- Reinforcement Learning: Commercial and Intellectual Value (7:47)
- Let's Connect
- Reinforcement Learning Simulation Packages in Python (5:58)
- Installing OpenAI Gym (gym[all]) on Linux, Windows and Mac (10:32)
- OpenAI Gym: How to Start an Environment and Visualize it (6:28)
- Coding Exercise: Set up the BipedalWalker-v3 environment
- OpenAI Gym: How to Observe the Environment (6:56)
- Quiz: Observation Elements in BipedalWalker-v3
- Coding Exercise: Interpret the Observation Space
- OpenAI Gym: How to Take Actions (7:30)
- Quiz: Action Elements in BipedalWalker-v3
- Coding Exercise: Taking Actions in BipedalWalker-v3
- OpenAI Gym: Rewards and Goals (7:00)
- Quiz: Real World Goals
- Coding Exercise: Reward for Falling Down in BipedalWalker-v3
- OpenAI Gym: Terminal States and Episodes (8:46)
- Quiz: Terminal State in Video Games
- Coding Exercise: Calculate Expected Cumulative Rewards per Episode
- How Reinforcement Learning Algorithms Work: A High Level Overview (9:32)
- Quiz: Deep Neural Nets in RL
- Which Reinforcement Learning Framework is the Best? (7:05)
- How to Install Ray RLlib (2:16)
- Ray RLlib: How to Use Deep RL Algorithms to Solve RL Problems (10:44)
- Coding Exercise: Teach a Robot How to Walk
- Ray RLlib: How to Visualize Results Using Tensorboard (8:12)
- Coding Exercise: Visualize Results from the BipedalWalker-v3 PPO Experiment
- Ray RLlib: How to Save a Trained Agent for Later Use (3:15)
- Coding Exercise: Save the Trained Robot
- Ray RLlib: How to Use and Record a Saved Agent (7:29)
- Coding Exercise: Create a Video of the Walking Robot
- How to Choose an Appropriate Deep RL Algorithm for Your Problem (6:15)
- Quiz: Algorithm Selection in the Inventory Management Problem
- How Was Your Experience?
- Where to Go from Here? (4:38)
Features
Easy to digest
Bite sized video lessons with no fluff (on an average 6 mins long and rarely over 10 mins).
The whole course can be completed in 4 hours (including exercises).
Easy to follow
All videos have closed captions.
Learn by doing
Video lessons and demonstrations are followed by coding exercises whenever possible.
Project based
The exercises are part of an overarching project, where we teach a robot how to walk. We will record a video of this agent at the end of the course, making it easy to share your new skills with others (if you wish).
Certificate of Completion
You will get a downloadable course completion certificate when you finish the course
Hi, I am Dibya, the instructor of this course 👋
- I am a Senior Python Developer, working closely with one of the biggest automotive companies in Germany.
- I organize the 3000 member Python Meetup community in Munich, Germany.
- I teach a DataCamp course on Unit Testing for Data Science, with 17500+ students.
- I have trained 300+ developers in the domain of Deep RL.
Let's get started 🚀
Have 4 hours of time and the curiosity to learn and apply Deep RL? Then let's get started.
You'll gain practical Deep RL skills using industry-leading tools with The Fast Deep RL Course.
If you have any questions, please drop me a message. I am looking forward to seeing you inside!
*free until 08.07.2022