Tutorials (RLDM 2017)

RLDM2017 will begin with 4 tutorials on Sunday, June 11

With permission of speakers tutorials will be recorded and made available online after the conference.

The program will consist, first, of a pair of basic tutorials, in two tracks, one (Joelle Pineau) intended to introduce the basics of computational reinforcement learning to an audience trained in psychology or neuroscience; the other (Anne Collins) to introduce the psychology and neuroscience of reinforcement learning to people with more background in AI.

The tracks will then merge for two advanced tutorials on state-of-the-art issues in either subfield; John Schulman on deep RL in AI and Matt Botvinick on understanding human and animal brains in terms of advanced and deep RL methods.


Sunday June 11, 2017- Tutorials

11:00-5:00pm- Registration desk open (Rackham Lobby)

12pm-5pm- Refreshments available in the East & West Conference Rooms

Track 1- Location: Rackham Assembly Hall.

1:00- 2:40pm Basic tutorial, (intro to AI/RL track): Joelle Pineau

Track 2 – Location: Rackham Amphitheatre.  Capacity 227 ppl

1:00- 2:40pm Basic tutorial (intro to Bio/psych track): Anne Collins

Session 2 – Location: Rackham Amphitheatre.  Capacity 227 ppl

2:45-4:10 Advanced tutorial, John Schulman: “Deep Reinforcement Learning Through Policy Optimization”
4:30-6:00 Advanced tutorial, Matthew Botvinick: “Deep RL, fast and slow”

6:00pm- 12:00am Private tent, bar and activities at Ann Arbor Summer Festival’s “Top of the Park”

Tutorial Information

Basics of Computational Reinforcement Learning
Joelle Pineau, McGill University

The problem of reinforcement learning is concerned with using experience gained through interacting with the world and evaluative feedback to improve a system’s ability to make a sequence of decisions. This tutorial will introduce the fundamental concepts and vocabulary that underlie this field of study. We will review key concepts (Bellman equation, exploration/exploitation, on/off-policy) and basic algorithms (value/policy iteration, actor-critic, policy search, deep Q networks). We will also discuss in what way reinforcement learning is different from other machine learning branches (e.g. supervised or unsupervised learning) and what type of problems should be tackled with reinforcement learning.

Joelle Pineau is an Associate Professor and William Dawson Scholar at McGill University where she co-directs the Reasoning and Learning Lab. Dr. Pineau’s research focuses on developing new models and algorithms for planning and learning in complex partially-observable domains. She also works on applying these algorithms to complex problems in robotics, health care, games and conversational agents. She serves on the editorial board of the Journal of Artificial Intelligence Research and the Journal of Machine Learning Research and is currently President-Elect of the International Machine Learning Society. She is a Senior Fellow of the Canadian Institute for Advanced Research and in 2016 was named a member of the College of New Scholars, Artists and Scientists by the Royal Society of Canada.

Anne Collins, University of California, Berkeley

John Schulman, OpenAI

John Schulman received his PhD in Computer Science from UC Berkeley where he co-taught the first class on deep reinforcement learning. He is now a research scientist at OpenAI where he focuses primarily on deep RL.

Matt Botvinick, DeepMind and Gatbsy Computational Neuroscience Unit, UCL