Neural Models for Part-Whole Hierarchies
Max Riesenhuber   Peter Dayan
NIPS 9, 661-667.
Abstract
We present a connectionist method for representing images that
explicitly addresses their hierarchical nature. It blends data from
neuroscience about whole-object viewpoint sensitive cells in
inferotemporal cortex (Logothetis, Pauls & Poggio, 1995) and
attentional basis-field modulation in V4 (Connor et al,
1996) with ideas about hierarchical descriptions based on
microfeatures (Hinton, 1981; Pollack, 1990). The resulting model makes
critical use of bottom-up and top-down pathways for analysis and
synthesis (Hinton et al, 1995). We illustrate the model
with a simple example of representing information about faces.
compressed postscript   pdf