University of Birmingham > Talks@bham > Artificial Intelligence and Natural Computation seminars > Vision as encoding, selection, and decoding: a view from the primary visual cortex

Vision as encoding, selection, and decoding: a view from the primary visual cortex

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Lars Kunze.

Host: Prof. Xin Yao, Prof. Uta Noppeney, Dr. Ulrik Beierholm

Abstract: One traditional view divides vision into low level, mid-level, and high level components; and another is Marr’s division of visual tasks into building a primal sketch, 2.5D sketch, and 3D model of the visual world. I propose to view vision as composed of encoding, selection, and decoding stages. Encoding samples and represents the visual input information in an optimal manner in the neural activities of, e.g., the retinal ganglion cells. Selection by attention picks a tiny fraction of the encoded input for deeper processing. Decoding concerns inferring the properties of a visual scene, e.g., the identity and motion of a 3D object, from the encoded and (mainly or exclusively) selected information. Unlike the traditional views, this proposed three-stage view of vision highlights the critical role of information selection. As in Marr’s approach, it makes explicit the task at each stage, putting the research questions into sharp focus. However, whereas Marr’s different stages differ mainly in the sophistication of the representation of the visual world, I argue that the three visual stages, encoding, selection, and decoding, solve fundamentally different problems. I will illustrate examples of these three stage through their manifestations in the primary visual cortex.

Short bio: Li Zhaoping is professor of computational neuroscience in the computer science department of University College London (UCL), and is the author of the textbook “Understanding vision: theory, models, and data” (Oxford University Press 2014). She obtained her Ph.D. in physics in 1989 from California Institute of Technology, and did her postdoctoral training in Institute for Advanced Study in Princeton and Rockefeller University in New York. She joined UCL in 1998 when she helped to found the Gatsby Computational Neuroscience Unit.

Speaker’s webpage: http://www.cs.ucl.ac.uk/staff/Zhaoping.Li/

This talk is part of the Artificial Intelligence and Natural Computation seminars series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

Talks@bham, University of Birmingham. Contact Us | Help and Documentation | Privacy and Publicity.
talks@bham is based on talks.cam from the University of Cambridge.