Nonlinear higher-order statistical decorrelation by volume-conserving neural architectures

Neural Networks

Volumn 8, Issue 4, 1995, Pages 525-535

  Deco, Gustavo ^a   Brauer, Wilfried ^a  

^a SIEMENS AG   (Germany)

Author keywords

Factorial learning; Nonlinear decorrelation; Volume conserving architectures

Indexed keywords

COMPUTER ARCHITECTURE; CORRELATION METHODS; INFORMATION THEORY; LEARNING SYSTEMS; MATHEMATICAL MODELS; MATHEMATICAL TRANSFORMATIONS; PROBABILITY; STATISTICAL METHODS; TENSORS;

BARLOW THEORY; FACTORIAL LEARNING; NEURAL ARCHITECTURES; NONLINEAR PRINCIPAL COMPONENT ANALYSIS; NONLINEAR STATISTICAL DECORRELATION;

NEURAL NETWORKS;

ARTICLE; COGNITION; COMPUTER MODEL; CONTROLLED STUDY; INFORMATION RETRIEVAL; LEARNING; PRIORITY JOURNAL;

EID: 0029013513     PISSN: 08936080     EISSN: None     Source Type: Journal    
DOI: 10.1016/0893-6080(94)00108-X     Document Type: Article

References (22)

1. Towards a theory of early visual processing
2. What does the retina know about natural scenes
3. Sensory mechanism, the reduction of redundancy, and intelligence
4. Unsupervised learning
5. Adaptation and decorrelation in the cortex
6. Finding minimum entropy codes
7. Nonlinear features extraction by redundancy reduction with stochastic neural networks
8. Product units A computationally powerful and biologically plausible extension to backpropagation networks
9. Adaptive network for optimal linear feature extraction
10. The anti-Hebb rule derived from information theory
11. Self-organization in a perceptual network
12. How to generate ordered maps by maximizing the mutual information between input and output signals
13. Local synaptic learning rules suffice to maximize mutual information in a linear network
14. Generalized linear features extraction: An information theory approach
15. Redundancy reduction as a strategy for unsupervised learning
16. Supervised factorial learning
17. A self-organization network for principal-component analysis
18. Learning higher order correlations