Schedule
Aug 04, 2001
Filtering and Optimal Decision Strategies for Sensorimotor Control
Dr. Bill Bialek
http://www.neci.nj.nec.com/neci-website/bios/bialek.html
What are the strategies used by animals to extract features from
their sensory field as a means to plan their motor output? We
examine the issue of coding under various behavior strategies. For
the case of feedback control in visually guided flight, is it
essential to maintain a high fidelity copy of the visual field as
a means to compare this with a desired set- point. This strategy
is illustrated for the case of neuron H1 in the fly, for which the
neuron is shown to minimize the error between an actual velocity
field and that encoded by a spike train. We further show how the
neuronal computation for velocity prediction, again using H1 as an
example, changes as a function of visual contrast. The consequence
of different coding strategies on motor control, in which at least
some information about the sensory field must be shed, is
discussed. Lastly, the issue of compression of sensory input and
recall in memory systems is discussed in terms of minimization
principles.
Audio for this talk requires sound hardware, and RealPlayer or RealAudio
by RealNetworks.
First part of lecture
Begin WebCam and audio: high bandwidth or medium bandwidth.
Or, begin audio only:
high bandwidth or low bandwidth.
(Or, right-click to download the whole audio file.)
Second part of lecture
Begin WebCam and audio: high bandwidth or medium bandwidth.
Or, begin audio only:
high bandwidth or low bandwidth.
(Or, right-click to download the whole audio file.)
Author entry (protected)