A layered recurrent neural network for feature grouping
We describe a recurrent network, the Competitive Layer Model (CLM) for feature grouping. The model uses a combination of cooperative and competitive interactions to partition a set of input features into salient groups whose number is only restricted by the available layers. We give analytic results on convergence and the attractor states of the model and present simulation results showing grouping by proximity and grouping by symmetry and good continuation.
Unable to display preview. Download preview PDF.
- 1.S. A. Ritz J. A. Anderson, J. W. Silverstein and R. S. Jones. Distinctive features, categorical perception, and probability learning: Some applications of a neural model. Psychological Review, 84:413–451, 1977.Google Scholar
- 2.R. Lippmann. An introduction to computing with neural nets. IEEE ASSP Mag., pages 4–22, 1987.Google Scholar
- 3.D. G. Luenberger. Introduction to dynamical systems: Theory, models, and applications. New York: Wiley, 1979.Google Scholar
- 4.Yeshurun H. W. Reisfeld, D. Context-free attentional operators: The generalized symmetry transform. Int. Journal of Computer Vision, 14:119–130, 1995.Google Scholar
- 5.H. Ritter. A spatial approach to feature linking. In Int. Neur. Netw. Conf. Paris, 1990.Google Scholar
- 6.R. Ritz, W. Gerstner, U. Fuentes, and J.L. van Hemmen. A biological motivated and analytically soluble model of collective oscillations in the cortex. II. Application to binding and pattern segmentation. Biol. Cybern., 71:349–358, 1994.Google Scholar
- 7.C. v.d. Malsburg. The correlation theory of brain function. Technical Report 81-2, MPI Göttingen, 1981.Google Scholar