Host: Darcy Kelley
Title: Neuroanatomy is Destiny: The Comparative Study of Brain Circuits and Implications for Machine Learning
Abstract: The study of intelligent behavior has traditionally started from a psychological-behavioral perspective. We argue that comparative neuroanatomy, taken together with the analysis of distributed algorithms for machine learning, provides an alternative route: the direct, empirical study of the circuit architectures comprising real and engineered brains. In this talk we will present comparative work analyzing mesoscale circuit anatomy in a rodent (mouse) and a primate (marmoset) brain using the first data set gathered using uniform methods across the two species. In the second part of the talk, theoretical analysis of overparameterized learning algorithms will be used to shed light on the similarities and differences between biological brains and contemporary " engineered brains".