Next Session Date: October 2nd, 2021, 9 AM PDT
Please sign up for roles as early as possible if you wish to present for that role to give your self space to prepare
A Template for different roles has been created and you can use this to prepare your presentation.
📅 Date
🗺️ Where
Google Meet link: https://meet.google.com/wao-jmds-rzw
MLC RL Reading Group Session #6 Saturday, October 2nd· 9:30 – 10:30 pm IST Google Meet joining info Video call link: https://meet.google.com/wao-jmds-rzw
Format
Paper Title
Reinforcement learning with unsupervised auxiliary tasks
https://arxiv.org/pdf/1611.05397.pdf
Blog Post :
https://deepmind.com/blog/article/reinforcement-learning-unsupervised-auxiliary-tasks
Abstract
Deep reinforcement learning agents have achieved state-of-the-art results by di- rectly maximising cumulative reward. However, environments contain a much wider variety of possible training signals. In this paper, we introduce an agent that also maximises many other pseudo-reward functions simultaneously by rein- forcement learning. All of these tasks share a common representation that, like unsupervised learning, continues to develop in the absence of extrinsic rewards. We also introduce a novel mechanism for focusing this representation upon ex- trinsic rewards, so that learning can rapidly adapt to the most relevant aspects of the actual task. Our agent significantly outperforms the previous state-of-the- art on Atari, averaging 880% expert human performance, and a challenging suite of first-person, three-dimensional Labyrinth tasks leading to a mean speedup in learning of 10× and averaging 87% expert human performance on Labyrinth.