Deep Reinforcement Learning & Control

Spring 2017
/ Machine Learning /

Pre-requisites: None
Co-requisites: None
Units: 12.0


This course will cover latest advances in Reinforcement Learning and Control, such as, deep Q learning, actor-critic methods, learning and planning, concurrent trajectory optimization and policy learning, inverse reinforcement learning, hierarchical reinforcement learning methods, forward predictive models, deep model predictive control, exploration strategies, adaptive control, applications to deep robotic learning. By the end of the course you should be able to: 1) code up a suitable reinforcement learning method in simulation or on robotic platform for a task 2) identify what are easy and hard problems in RL and learning for robotics The course will have a final project which will involve design of a reinforcement learning method in simulation or robotic platform. The homeworks will be in OpenAI gym. Pre-requisites: Students should have a basic background in algorithms, linear algebra, machine Learning, deep learning.


['Salakhutdinov, Ruslan', 'Fragkiadaki, Aikaterini']
today Monday, Wednesday
access_time From 03:00PM to 04:20PM
explore GHC 4401

['Fragkiadaki, Aikaterini', 'Salakhutdinov, Ruslan']
explore DNM DNM


cmu 10-703 - CMU Coursefind