Andrew G. Barto is a professor emeritus of computer science at the University of Massachusetts Amherst, where he co-founded and co-directed the Autonomous Learning Laboratory and served as department chair. His research connects reinforcement learning to neuroscience, with a focus on reward signals in the brain and biologically plausible methods for learning in artificial neural networks. He joined UMass in 1977 and retired in 2012.
Barto is one of the two founders of modern reinforcement learning, the branch of machine learning in which an agent learns by trial and error from rewards. Beginning in the 1980s, Barto and his former graduate student Richard Sutton introduced the main ideas, built the mathematical foundations, and developed the core algorithms of the field. The two co-authored “Reinforcement Learning: An Introduction,” the standard textbook on the subject. For this body of work they shared the 2024 ACM Turing Award.
For the library’s reader, Barto is the other name behind the reinforcement-learning entries that more often mention Sutton. The ideas the two developed, learning to make good sequences of decisions when rewards are delayed, are what later powered game-playing systems and the human-feedback tuning of AI assistants. Barto’s particular contribution kept one foot in the brain: his work treated reinforcement learning not just as an engineering trick but as a candidate theory of how real nervous systems learn.