Short bio: Pietro Perona is interested in the computational foundations of vision in both humans and machines. He has worked on partial differential equations for image processing (anisotropic diffusion), texture discrimination, edge detection, motion analysis and 3D reconstruction. He is currently studying visual recognition and the analysis of biological motion. Dr. Perona graduated in Electrical Engineering from the University of Padova in 1985, and received a PhD in Electrical Engineering and Computer Science from the University of California at Berkeley in 1990. After a year as a post-doctoral Fellow at MIT, he joined the faculty of the California Institute of Technology where he is currently Allen E. Puckett Professor of Electrical Engineering and Executive Officer of the Computation and Neural Systems option.

Abstract
One of the most remarkable functions of our senses is categorization, which allows us to predict the properties of objects and situations we have never encountered before. Categories are not innate, but rather learnt through experience, with very little supervision, and amidst overwhelming amounts of clutter and noise. How does this happen? I will propose a taxonomy of visual categorization and explore approaches and experiments that give an `existence proof' of solutions to this problem. I will highlight applications and open questions.