Model of competitive learning based upon a generalized energy function

✍ Scribed by Tadashi Masuda

• Book ID: 104348693
• Publisher: Elsevier Science
• Year: 1993
• Tongue: English
• Weight: 798 KB
• Volume: 6
• Category: Article
• ISSN: 0893-6080
• DOI: 10.1016/s0893-6080(09)80021-x

✦ Synopsis

A neuralnetworkmodel of competitive learning wasproposed. In this model, output cells in thenetwork were self-organized to represent the distribution of input pattern vectors. The self-organization was based upon a generalized energyfunction. The network was mathematically proved to converge to the global minimum of the energy function when the number of output cells is the same as that of input patterns. In this global minimum, a one-to-one correspondence was established between inputpatterns and output cells, and an output cell responded exclusively to its corresponding input pattern. The model was compared with conventional models of competitive learning orfeature detection. Typical behavior of the network was demonstrated by computer simulation, which included the case of clustered inputpatterns.