Faculty Spotlight: Payam Piray

The Piray Lab focuses on how people make decisions in noisy environments through computational models using reinforcement learning and Bayesian machine learning.

The star of the show this week is Prof. Payam Piray, who worked with Nathaniel Daw as a Postdoc at Princeton. Prof. Piray’s research focuses on how we make decisions in an uncertain world. Below are two of his ongoing projects.

Linear RL: When solving a planning problem, the brain often resorts to adjusting previous solutions to similar problems, rather than starting from zero. AI systems can’t yet do this as efficiently as the brain. Reinforcement learning (RL) models of decision making are either too computationally complex or don’t translate well to the real world. Piray suggests a linear RL model which reuses previous computations in a way that’s much closer to what your brain does. This creates a much more flexible model that is supported by data collected from grid and border cells in the brain.

Stochasticity and Volatility: Here’s another problem the brain solves better than we do: adaptive learning requires one to differentiate between two types of uncertainty: volatility (speed of change) and stochasticity (randomness of change). Piray suggests that most existing systems focus on learning one factor by itself, disregarding the other, causing faulty results. To fix this, he’s developed a way of learning both factors simultaneously, allowing for a better approximation of what the brain actually does.

Professor Piray is currently looking for interested PhD students and is on leave, but will be back at the start of November. You may reach him at piray@usc.edu.

Thank you for reading. Reminder: Byte Sized is open to everyone! Feel free to submit your piece.

All submissions to berkan@usc.edu with the header “Byte Sized Submission” please. Thank you!