Gerald J Tesauro, IBM TJ Watson Research Center
Balaraman Ravindran, Indian Institute of Technology Madras
Sridhar Mahadevan, University of Massachusetts Amherst
Camera ready submission due:
Workshop: July 11, 2016 (Time: 9 AM - 5:30 PM EDT)
Venue: New York Hilton Midtown, Room Concourse A
There has been a resurgence of Neural Networks in the name of Deep Learning and recent researches have demonstrated the use of neural networks as powerful Representation Learning algorithms. Reinforcement Learning is one field that has been using neural networks as function approximators (aka representation learners) for more than two decades. In this workshop, we will focus on various ways in which representation learning, and reinforcement learning interact. Very recently, Deep Neural Network based representations were used along with Q-Learning to learn human level control to play video games. In the other direction, REINFORCE is being used in several deep learning models to learn complex tasks like image classification, and image description. It is very exciting to see that both the fields contribute to each other. This workshop will focus on Deep Reinforcement Learning, where DL is helpful in learning representations for RL, and Reinforced Deep Learning, where RL is helpful in training Deep Neural Networks. The aim of this workshop is to bring researchers from both the fields together and discuss new challenging applications which requires both Deep Learning and Reinforcement Learning.
Sarath Chandar, University of Montreal