Abstract

Meeting ID: 982 7309 9888
Passcode: 582157

Convolutional neural networks (CNNs) have been proposed as computational models for (rapid) human object recognition and the (feedforward-component) of the primate ventral stream. The usefulness of CNNs as such models obviously depends on the degree of similarity they share with human visual processing. In my talk I will discuss major differences between human vision and currently used standard CNNs in terms of their robustness to image distortions (out-of-distribution generalisation) and texture bias. In addition, I will argue for the importance of comparing decision makers on a trial-by-trial basis rather than simply on aggregated performance if we want to make claims about similar strategies or internal representations.