University of Cambridge > Talks.cam > Microsoft Research Machine Learning and Perception Seminars > Reverse Engineering the Human Visual System with Networks of Spiking Neurons

Reverse Engineering the Human Visual System with Networks of Spiking Neurons

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

If you have a question about this talk, please contact Oliver Williams.

Can we use information about the way in which biological visual systems function to reverse engineer the brain? As long ago as 1989, I had argued that the speed of processing achieved by the primate visual system can be used to constrain processing algorithms and architectures. Recently, we have found evidence that the human visual system can detect the presence of important visual patterns such as faces and animals in cluttered visual environments in under 100 ms. Such data implies that a lot can be done with just a single feed-forward pass through a series of hierarchically organised processing stages. This conjecture seems to be confirmed by results from a number of recent artificial vision systems that achieve impressive levels of performance in scene categorisation tasks using very similar strategies, and this leads to optimism that we could be heading for the long-awaited convergence between biological and machine vision. But there is another feature of biological vision that is absent from nearly all machine vision systems – biological vision does processing with spiking neurons. I will argue that by using the wave of spikes that is initiated by the presentation of a visual stimulus, and specifically the fact that the most strongly activated neurons tend to fire first, the visual system can achieve processing speeds that would be difficult if not impossible to achieve using more conventional techniques. Such ideas can be used to develop powerful image processing methods that can be even more efficient that human vision.

This talk is part of the Microsoft Research Machine Learning and Perception Seminars series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2021 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity