THIS PAGE INCORPORATES CASCADING STYLE SHEET! TO VIEW PAGE CORRECTLY PLEASE UPDATE TO A CSS COMPLIANT BROWSER
 |
Localized Receptive FieldsMay Mediate Trasformation-Invariant Recognition in the Visual Cortex |
pdf 154 KB
ps.gz 120 KB
ps 730 KB
@TechReport{Rao:TR97b,
author = {Rajesh P.N. Rao and Dana H. Ballard},
title = {Localized Receptive FieldsMay Mediate Trasformation-Invariant Recognition in the Visual Cortex},
institution = {CS Dept. University of Rochester},
year = {1997},
number = {97.2},
address = {Rochester NY},
month = {May},
abstract = {Neurons in the visual cortex are known to possess localized, oriented receptive fields. It
has previously been suggested that these distinctive properties may reflect an efficient image
encoding strategy based on maximizing the sparseness of the distribution of output neuronal
activities or alternately, extracting the independent components of natural image ensembles.
Here, we show that a relatively simple neural solution to the problem of transformationinvariant
visual recognition also causes localized, oriented receptive fields to be learned from natural
images. These receptive fields, which code for various transformations in the image plane,
allow a pair of cooperating neural networks, one estimating object identity (``what'') and the
other estimating object transformations (``where''), to simultaneously recognize an object and
estimate its pose by jointly maximizing the a posteriori probability of generating the observed
visual data. We provide experimental results demonstrating the ability of these networks to
factor retinal stimuli into objectcentered features and objectinvariant transformations. The
resulting neuronal architecture suggests concrete computational roles for the neuroanatomical
connections known to exist between the dorsal and ventral visual pathways. }
}