Nunug

Topic: Bootstrapping Intelligence: Getting Started with Deep Reinforcement Learning

Presenter: Bruce Nielson

Date: Thursday, January 13, 2022

Time: 6:30:00 PM -

Location: https://zoom.us/j/97500146056?pwd=bXVGeWZVS0c3REVSSlBZZTVtTndyZz09

Reinforcement Learning has created a lot of excitement with its success in Alpha Go beating the world Go champion, Lee Sedol. In this presentation, I’ll introduce you to Q Learning and how to convert a Q Learning algorithm into a Deep Reinforcement Learning algorithm to solve one of Open AI’s Open Gym problems: Lunar Lander.Deep Reinforcement Learning replaces a Q Learning Q-table (a representation of every possible configuration of the problem you wish to solve) with a Deep Learning Neural Network. It works off an idea that is a bit hard to believe it even works – letting the Neural Net predict its own ground truth for training. By bootstrapping itself and letting rewards from the real world slowly guide it, it can successfully solve Open Gym games like Lunar Lander.

About the presenter: