We’ve put together a series of Training Videos to teach customers about reinforcement learning and The Bonsai Platform. Check out Video 1 to get started with an introduction to types of machine learning.
Today we have uploaded the fifth video in a series of training videos to get new customers quickly up to speed on the Bonsai platform, the Inkling programming language, and reinforcement learning. You might be wondering, “where are videos three and four?” “Why isn’t this one the third video in the series?” The truth is, this training series all started from giving this presentation to an early access customer, deciding to turn it into a video, and then planning out a curriculum to set the stage for it. Videos three and four will follow this one.
We realized that other companies might not have such a strong grasp on machine learning as our initial customers did, so we set out to break up the on-site training into smaller, easily digestible videos for you to watch in your leisure. Today’s video is Challenges and Strategies in Reinforcement Learning (RL) and is broken up into two parts; an overview of three main challenges in RL, and general strategies to make RL problems easier to solve.
The video assumes that you already have a general understanding of reinforcement learning, which you can get from watching the previous videos in this series. You’ll learn about challenges with dynamic programming problems, reward functions, and scalability of learning. Turning to solutions, you’ll then learn some strategies to overcome these challenges, such as shaping, curriculum learning, apprenticeship learning, and using building blocks. Check out the full video below.
For more information about how Bonsai uses reinforcement learning you can read our blog on Mark Hammond’s Deep Reinforcement Learning presentation at GTC this year.