BEGIN:VCALENDAR VERSION:2.0 CALSCALE:GREGORIAN PRODID:iCalendar-Ruby BEGIN:VEVENT CATEGORIES:Concerts & Performances DESCRIPTION:Public Dissertation Defense by Indupama Herath\, March 1\n\nDoc toral candidate Indupama Herath discusses her dissertation\, "Multivariate Regression using Neural Networks and Sums of Separable Functions\," on Tues day March 1\, at 9 a.m. via Teams.\n\nJoin the meeting via Teams. \n\nHerat h is a graduate student in Mathematics in the College of Arts & Sciences.\n \n \n\nAbstract: Currently\, artificial neural networks are the most popula r approach to machine learning problems such as high-dimensional multivaria te regression. Methods using sums of separable functions are designed to re present functions in high dimensions and can be applied to high-dimensional multivariate regression. Here we compare the ability of these two methods to approximate function spaces in order to assess their relative expressive power.\n\n \n\nWe show that translation of approximation results from one method to the other is possible under certain conditions. We identify gener al approximation schemes in both the single-layer and deep-layer settings t hat apply to both methods for approximating certain function classes. In pa rticular\, we show that sums of separable functions give the same error rat es as neural networks for function classes such as Barron's functions and b and-limited functions. Inspired by deep neural networks\, we also introduce deep layer sums of separable functions that shows similar results as deep neural networks for functions with compositional structure. DTEND:20220301T150000Z DTSTAMP:20241124T070321Z DTSTART:20220301T140000Z LOCATION: SEQUENCE:0 SUMMARY:Public Dissertation Defense by Indupama Herath\, March 1 UID:tag:localist.com\,2008:EventInstance_39245597788725 URL:https://calendar.ohio.edu/event/public_dissertation_defense_by_indupama _herath_march_1 END:VEVENT END:VCALENDAR