UFII Fellows Journal Club Seminar Series
“SINs: Simulation Inversion Networks for Automated Fitting of Mechanistic Models”
by Konstantina Sokratous
UFII Fellow and Ph.D. Candidate, Behavioral and Cognitive Neuroscience, Department of Psychology
Monday, April 11, 2022 HYBRID: In-person with lunch at 11:30AM; live and zoom talk at noon
Mathematical and computational modeling are crucial tools when unraveling the properties of complex and dynamic systems, such as the human brain. However, the usability of generative models is severely limited by the intractable nature of their likelihood function or its inability to computed in polynomial time, rendering traditional methods for parameter estimation inapplicable and approximation methods impractical. Since simulating a process from these models remains an option even without a solution to the probability density, we present how artificial neural networks can overcome computation difficulties allowing for automated model fitting and instantaneous parameter extraction. For illustration purposes, we provide an example of our method on a cognitive stochastic model of valuation.
Konstantina Sokratous is a second-year graduate student in the Behavioral and Cognitive Neuroscience area, working with Dr. Peter Kvam. She received a UF Informatics Institute Graduate Fellowship to develop machine learning approaches to model fitting, and has been working on projects related to modeling cognitive processes underlying anticipatory decision-making and timing, behavior on risky and impulsive prospects, pricing of delayed option, and decisions among alternatives that change over time. She is also interested in new methods for studying psychology and cognition, along with applying computational methods to the study of neural and evolutionary processes.
Monday, April 11, 2022
12:00PM – 1:00PM
Lunch will serve at 11:30AM
When you register below, please indicate if you will be attending in-person or via zoom. Zoom invite will be sent out the day before the talk.