A new competitive learning (CL) algorithm with k-winners-take-all activation is presented. The k winning neurons for updating are those best matching the input vector in the wavelet domain with subspace search. Simulation results show that the algorithm gives a better performance than that of the traditional CL algorithm while requiring much less computational time.
|Number of pages||2|
|Publication status||Published - 1998 Jun 11|
ASJC Scopus subject areas
- Electrical and Electronic Engineering