Reinforcement learning is learning what to do -- how to map situations into actions -- so as to maximize a numerical reward signal. The learner is not told which actions to take, as in most forms of machine learning, but instead must discover which actions yield the most reward by trying them.  Over 30 years ago, the computational study of reinforcement learning began in Professor Andrew Barto’s lab at UMass Amherst. It has since blossomed into an active and diverse area of research, combining machine learning, dynamic programming and Markov decision processes, cognitive science, neuroscience, control theory, and many application areas.

PICTURES from the workshop can be seen here

Organizers: Satinder Singh, Univ. of Michigan Computer Science: baveja@cs.umich.edu

Sridhar Mahadevan: UMass, Amherst, Computer Science: mahadeva@cs.umass.edu