Artificial vision by multi-layered neural networks: Neocognitron and its advances

• Published in: Neural networks Vol. 37; pp. 103 - 119
• Main Author: Fukushima, Kunihiko
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.01.2013; Elsevier
• Subjects: Animals; Applied sciences; Artificial Intelligence; Artificial vision; Attention - physiology; Biological and medical sciences; Bottom-up and top-down; Cognition - physiology; Computer science; control theory; systems; Connectionism. Neural networks; Exact sciences and technology; Fundamental and applied biological sciences. Psychology; Hierarchical network; Humans; Language; Learning - physiology; Mammals; Modeling neural networks; Neocognitron; Neural Networks (Computer); Pattern Recognition, Visual - physiology; Pattern recognition. Digital image processing. Computational geometry; Psychology. Psychoanalysis. Psychiatry; Psychology. Psychophysiology; Vision, Ocular - physiology; Visual Pathways - physiology; Visual Perception - physiology; Neocognitron; Bottom-up and top-down; Modeling neural networks; Hierarchical network; Artificial vision; Occlusion; Image recognition; Top down method; Neural network; Modeling; Visual system; Selective attention; Vertebrata; Mammalia; Completeness; Bottom up method; Hierarchical system; Occultation; Multilayer network
• ISSN: 0893-6080; 1879-2782
• DOI: 10.1016/j.neunet.2012.09.016

The neocognitron is a neural network model proposed by Fukushima (1980). Its architecture was suggested by neurophysiological findings on the visual systems of mammals. It is a hierarchical multi-layered network. It acquires the ability to robustly recognize visual patterns through learning. Although the neocognitron has a long history, modifications of the network to improve its performance are still going on. For example, a recent neocognitron uses a new learning rule, named add-if-silent, which makes the learning process much simpler and more stable. Nevertheless, a high recognition rate can be kept with a smaller scale of the network. Referring to the history of the neocognitron, this paper discusses recent advances in the neocognitron. We also show that various new functions can be realized by, for example, introducing top-down connections to the neocognitron: mechanism of selective attention, recognition and completion of partly occluded patterns, restoring occluded contours, and so on.