6.884 Computational Sensorimotor Learning


Graduate Level
Units: 3-0-9
Prereqs: Permission of Instructor
Instructor: Professor Pulkit Agrawal
Schedule: Tue/Thu 2:30pm - 4:00pm, Room 2-105
This subject qualifies as an Artificial Intelligence concentration subject. Introduces the fundamental algorithmic approaches for constructing agents that learn to act in their environment from raw sensory observations. Topics include imitation learning, observation learning, self-supervised learning, reinforcement learning, inverse reinforcement learning, model learning from raw sensory observations. The course will also provide an overview of practical learning based approached used for navigation and robotic manipulation. A significant portion of the course will be devoted to reading research papers and working on a project.
Previous work with Machine Learning and Deep Learning is recommended.