Abstract: Mental rotation denotes our ability to manipulate representations of two-dimensional or three-dimensional objects mentally, which allows us for example to recognise an object seen from an unusual point of view. Deep networks lack such ability, and are indeed less robust than the human visual system to recognise objects in unusual poses. In this talk, I will discuss our ongoing efforts to understand what happens in the brain when it is performing mental rotation, and how similar operations could be implemented in deep learning.

Speakers:  Stephane Deny is assistant professor in the Departments of Neuroscience and Biomedical Engineering and of Computer Science at Aalto University. Before that, he did my PhD in Paris at the Vision Institute in computational neuroscience, and two postdocs respectively at Stanford in theoretical neuroscience and at Facebook AI Research Lab in machine learning. He is interested both in making advances in our understanding of the brain and advances in machine learning.

Affiliation: Aalto University

Place of Seminar:  Kumpula exactum D122 (in person) & zoom ( Meeting ID: 640 5738 7231 ; Passcode: 825217)