Skip to main content

JOB TALK: Machine Learning Principles Underlying Rational Behavior

Online Event
Gergo Orban
Friday, November 13, 2020, 11:00 am – 12:00 pm

Please register for this event under the link provided on the right. We will send the link to registered attendees 1 hour before the job talk starts.

ABSTRACT / In recent years machine learning has transcended a number of disciplines. A particularly enticing application of machine learning concerns human behavior since it allows the formalization of human judgements in terms of rational behavior. Analogous challenges in machine learning and biological intelligence inspire an intense cross-talk between these two disciplines and allow both the application of most recent advances in machine learning to understand how human judgements are shaped and provide inspirations for addressing challenges in machine intelligence through our extensive knowledge on human cognition. In this talk I will adopt a Bayesian probabilistic modelling framework, and in particular generative models to investigate a number of central topics in machine learning, such as model selection, lossy compression, non-parametric Bayesian inference, stochastic sampling, and hierarchical inference and their application to a wide range of cognitive phenomena. I will argue that probabilistic generative models are key to provide normative arguments of the structure in both intra-subject and inter-subject variance in phenomena relating to cognitive processes.

BIO / Gergő Orbán is head of the Computational Systems Neuroscience Lab at the Department of Computational Sciences, Wigner Research Center for Physics, Budapest. His research focusses on using machine learning ideas to address problems in cognitive science and neuroscience. He seeks to capitalize on the unprecedented opportunities offered by the fast-paced advances in machine learning and artificial intelligence to address complex, high dimensional and structured data. In particular, inspired by the inherent uncertainty that biological agents are faced with when dealing with a complex, yet limited and noisy data he seeks how probabilistic computations could support their decisions and looks for the signatures of these computations in behavioral and neural data. Gergő graduated from the Eötvös University with a degree in physics and was awarded PhD at the same university pursuing research in computational neuroscience. He was a postdoc with Eörs Szathmáry at the Collegium Budapest, Institute for Advanced Study, and later he was a Swartz postdoctoral at the Volen Center for Complex Systems, Brandeis University and a Marie Curie Fellow at the Department of Engineering, University of Cambridge. Gergő has returned to Budapest with a Lendület Young Investigator Award to establish his lab at the Wigner Center.