PHYSICS
AND ASTRONOMY
Science
at the Edge
Friday,
April 23, 2010
11:30a
1400
Biomedical & Physical Sciences Bldg
Refreshments
at 11:15a in 1400 BPS Bldg
Tatyana
Sharpee
Salk
Institute for Biological Studies
“Maximally
informative input/output functions in biological networks”
Many
organisms rely on complex biological networks both within and between cells to
process information about their environments. As such, performance of
these networks can be quantified using the tools of information theory.
In this talk I will describe an approach for finding input/output functions for
notes in the network that would allow the network to convey the maximal amount
of information about inputs drawn a given probability distribution. In
particular, optimal solutions will be compared for Gaussian and more
naturalistic, non-Gaussian distributions that take into account large
deviations common in the natural environment. This comparison reveals the
expected differences in the correlation strength between nodes as well as in
the sensitivity to multiple input dimensions for individual nodes.
Generally, the strength of interaction between any two nodes turns out to be
related to the presence of cusps in the input/output functions, with zero
interactions corresponding to the absence of cusps. This last property
allows one to compute the strength of interaction between the observed and
unobserved notes in the situation where the output of only some notes in a
network can be monitored. I will conclude with applications of these
results to retinal circuits and describe new algorithms that could improve the
resolutions of retinal implants.
***To
make an appointment to meet with the speaker, contact Prof. Lisa Lapidus ([log in to unmask])***
Kim Crosslan
Undergraduate Secretary
Dept. Physics & Astronomy
Michigan State University
1312 Biomedical & Physical Sciences
East Lansing, MI 48824
517-884-5531