Semilinear Predictability Minimization Produces Well-Known Feature Detectors
- 1 May 1996
- journal article
- Published by MIT Press in Neural Computation
- Vol. 8 (4) , 773-786
- https://doi.org/10.1162/neco.1996.8.4.773
Abstract
Predictability minimization (PM—Schmidhuber 1992) exhibits various intuitive and theoretical advantages over many other methods for unsupervised redundancy reduction. So far, however, there have not been any serious practical applications of PM. In this paper, we apply semilinear PM to static real world images and find that without a teacher and without any significant preprocessing, the system automatically learns to generate distributed representations based on well-known feature detectors, such as orientation-sensitive edge detectors and off-center–on-surround detectors, thus extracting simple features related to those considered useful for image preprocessing and compression.Keywords
This publication has 9 references indexed in Scilit:
- What Is the Goal of Sensory Coding?Neural Computation, 1994
- Learning Factorial Codes by Predictability MinimizationNeural Computation, 1992
- Understanding Retinal Color Coding from First PrinciplesNeural Computation, 1992
- Forming sparse representations by local anti-Hebbian learningBiological Cybernetics, 1990
- Analysis of Linsker's Simulations of Hebbian RulesNeural Computation, 1990
- A Self-Organizing Network for Principal-Component AnalysisEurophysics Letters, 1989
- Finding Minimum Entropy CodesNeural Computation, 1989
- From basic network principles to neural architecture: emergence of orientation columns.Proceedings of the National Academy of Sciences, 1986
- From basic network principles to neural architecture: emergence of orientation-selective cells.Proceedings of the National Academy of Sciences, 1986