Timezone: »

The Challenges of Real World Reinforcement Learning
Daniel Mankowitz · Gabriel Dulac-Arnold · Shie Mannor · Omer Gottesman · Anusha Nagabandi · Doina Precup · Timothy A Mann · Gabriel Dulac-Arnold

Sat Dec 12 08:30 AM -- 07:30 PM (PST) @ None
Event URL: https://sites.google.com/view/neurips2020rwrl »

Reinforcement Learning (RL) has had numerous successes in recent years in solving complex problem domains. However, this progress has been largely limited to domains where a simulator is available or the real environment is quick and easy to access. This is one of a number of challenges that are bottlenecks to deploying RL agents on real-world systems. Two recent papers identify nine important challenges that, if solved, will take a big step towards enabling RL agents to be deployed to real-world systems (Dulac et. al. 2019, 2020).The goals of this workshop are four-fold: (1) Providing a forum for researchers in academia, industry researchers as well as industry practitioners from diverse backgrounds to discuss the challenges faced in real-world systems; (2) discuss and prioritize the nine research challenges. This includes determining which challenges we should focus on next, whether any new challenges should be added to the list or existing ones removed from this list; (3) Discuss problem formulations for the various challenges and critique these formulations or develop new ones. This is especially important for more abstract challenges such as explainability. We should also be asking ourselves whether the current Markov Decision Process (MDP) formulation is sufficient for solving these problems or whether modifications need to be made. (4) Discuss approaches to solving combinations of these challenges.

Sat 8:30 a.m. - 8:40 a.m.
Introduction and Overview (Introduction) Daniel Mankowitz, Gabe Dulac-Arnold
Sat 8:40 a.m. - 9:20 a.m.

Real World RL Challenges

Aviv Tamar
Sat 9:20 a.m. - 10:00 a.m.

More practical Batch Offline Reinforcement Learning

Emma Brunskill
Sat 10:00 a.m. - 10:40 a.m.

Challenges for RL in Robotics

Jost Tobias Springenberg
Sat 10:40 a.m. - 11:20 a.m.
Mini-panel discussion 1 - Bridging the gap between theory and practice (Discussion Panel)
Aviv Tamar, Emma Brunskill, Jost Tobias Springenberg, Omer Gottesman, Daniel Mankowitz
Sat 11:20 a.m. - 11:50 a.m.

You can now chat to the paper authors by clicking the above Gather.town link

Links to individual poster presentations can be found here: https://sites.google.com/corp/view/neurips2020rwrl#h.ey6lwdtrdt7c

Sat 11:50 a.m. - 12:30 p.m.

Challenges of Model-based Inverse Reinforcement Learning

Sat 12:30 p.m. - 1:10 p.m.

Boston Dynamics

Sat 1:10 p.m. - 1:50 p.m.

The following speakers that will be at this event do not have Neurips profiles: Franziska Meier - fmeier@fb.com Marc Reibert - marcraibert@bostondynamics.com Scott Kuindersma - skuindersma@bostondynamics.com

Franziska Meier, Gabriel Dulac-Arnold, Shie Mannor, Timothy A Mann
Sat 1:50 p.m. - 3:20 p.m.

Enjoy your lunch break.

If you intend to attend the 3rd mini-panel session, we encourage you to watch the talks of Anca Dragan and Angela Schoellig during lunch as their keynote talks will only occur after the mini-panel session. Thus, if you want to ask them questions, please take the time to watch the talks now.

Sat 3:20 p.m. - 4:00 p.m.

We have 4 spotlight talks. These talks can be found at the following link: https://sites.google.com/corp/view/neurips2020rwrl#h.9w5kdo7eecim

Sat 4:00 p.m. - 4:40 p.m.

Applying RL to Ecosystem Management: Lessons Learned

Tom Dietterich
Sat 4:40 p.m. - 5:20 p.m.

Reinforcement Learning for Real Robots

Chelsea Finn
Sat 5:20 p.m. - 6:00 p.m.
Mini-panel discussion 3 - Prioritizing Real World RL Challenges (Discussion Panel)
Chelsea Finn, Tom Dietterich, Angela Schoellig, Anca Dragan, Anusha Nagabandi, Doina Precup
Sat 6:00 p.m. - 6:30 p.m.

You can now chat to the paper authors by clicking the above Gather.town link

Links to individual poster presentations can be found here: https://sites.google.com/corp/view/neurips2020rwrl#h.ey6lwdtrdt7c

Sat 6:30 p.m. - 7:10 p.m.

Machine Learning for Safety-Critical Robotics Applications

Angela Schoellig
Sat 7:10 p.m. - 7:50 p.m.

Reinforcement Learning that optimizes what people really want

Anca Dragan

Author Information

Daniel Mankowitz (DeepMind)
Gabriel Dulac-Arnold (Google Research)
Shie Mannor (Technion)
Omer Gottesman (Harvard)
Anusha Nagabandi (UC Berkeley)
Doina Precup (DeepMind)
Timothy A Mann (DeepMind)
Gabe Dulac-Arnold (Google Research)

More from the Same Authors