Chair: Andreas Cap
Abstract: The ubiquitous use of neural networks in any area which is at least mildly related to machine learning has tempted a lot of innocent mathematicians into trying to find a theoretical explanation for their observed success. While the outcome of this perilous endeavour yet remains shrouded in mystery, neural networks do certainly have a number of intriguing properties rendering their study not entirely futile. In this presentation I will attempt to convey some of their more alluring approximation theoretic aspects which have kept me entertained during my PhD studies.
Join Zoom Meeting: univienna.zoom.us/j/94428094604
Meeting ID: 944 2809 4604
Passcode: 83495