Resources
Blog posts, lectures, recorded talks, slides, and software.
Posts
-
Plan2Explore: Active Model-Building for Self-Supervised Visual Reinforcement Learning
Berkeley AI Research Blog
Oct 2020
Software
-
DreamerV2
Mastering Atari with Discrete World Models. Also for continuous control.
? Github stars
Mar 2021
-
BatchPPO
Simple parallel RL by batched action computation and separate env processes.
? Github stars
Oct 2017
Lectures
-
Learning World Models For Planning
Guest lecture (CSC2542)
University of Toronto
Mar 2021
-
Variational Inference in TensorFlow
Guest lecture (CS 20)
Stanford University
Feb 2018
-
Character-Level Language Modeling using Recurrent Neural Networks
Tutorial, 300 PhD attendees
PhD Intern Research Conference Workshop
Jul 2017
-
TensorFlow for Deep Reinforcement Learning
Guest lecture (CS 20SI)
Stanford University
Mar 2017
Orals
-
Evaluating Agents without Rewards
Contributed talk
NeurIPS Bio and Artificial RL Workshop
Dec 2020
-
Planning to Explore via Self-Supervised World Models
Oral presentation
ICML
Jun 2020
-
Dream to Control: Learning Behaviors by Latent Imagination
Oral presentation
ICLR
Apr 2020
-
Dream to Control: Learning Behaviors by Latent Imagination
Contributed talk
NeurIPS Deep RL Workshop
Dec 2019
-
Learning Latent Dynamics for Planning from Pixels
Oral presentation
ICML
Jun 2019
-
Sample-Efficient Reinforcement Learning with Stochastic Ensemble Value Expansion
Long oral presentation
NeurIPS
Dec 2019
Talks
-
General Infomax Agents through World Models
Invited talk
ICAPS 2021 AI Planning and Reinforcement Learning Workshop
Aug 2021
-
General Infomax Agents through World Models
Invited talk and panel discussion
ICML 2021 Unsupervised Reinforcement Learning Workshop
Jul 2021
-
Unsupervised Intelligent Agents
Invited talkm
ICLR 2021 Self-Supervised Reinforcement Learning Workshop
May 2021
-
Unsupervised Intelligent Agents
Invited talk and panel discussion
ICLR 2021 Never Ending Reinforcement Learning Workshop
May 2021
-
Action and Perception as Divergence Minimization
Invited talk
Deep Mind, Embodied Intelligence Group
Oct 2020
-
Action and Perception as Divergence Minimization
Invited talk
Deep Mind, Montreal Team
Oct 2020
-
Planning to Explore via Self-Supervised World Models
Invited talk
Intel AI Labs
Jul 2020
-
TalkRL: The Reinforcement Learning Podcast
TalkRL Podcast
May 2020
-
Efficient Visual Control with Learned Latent Dynamics
Invited talk
Facebook AI Research (FAIR)
Nov 2019
-
Planning and Active Inference using Expressive World Models
Invited talk
UCL Wellcome Centre for Human Neuroimaging (FIL)
Jul 2019
-
Learning Latent Dynamics for Planning from Pixels
Invited talk
DeepMind
Jun 2019
-
Planning and Exploration using Latent Dynamics Models
Invited talk
Berkeley AI Research (BAIR)
Jun 2019
-
Amortized Active Inference
Invited talk
UCL Wellcome Centre for Human Neuroimaging (FIL)
Jun 2018
-
Learning Hierarchical Information Flow with Recurrent Neural Modules
Invited talk
IBM Research
Feb 2018