OSNet: A neural network implementation of order statistic filters

TitleOSNet: A neural network implementation of order statistic filters
Publication TypeConference Paper
Year of Publication1991
AuthorsShi, P., and R. K. Ward
Conference NameCommunications, Computers and Signal Processing, 1991., IEEE Pacific Rim Conference on
Pagination453 -456 vol.2
Date Publishedmay.
Keywordsadaptive filters, adaptive OSNet, digital filters, input array, neural nets, neural network model, neurons, order statistic filters, order statistic network, OSNet, selection networks, total processing time

A neural network model, called OSNet (order statistic network), which finds the kth largest element in an array of integers is proposed. There are four layers of neurons in OSNet, thus the total processing time is four times the processing time of one single neuron. As the number of the elements in the input array increases, only the number of neurons in each layer increases. Therefore, the processing time of OSNet is constant irrespective of the number of elements in the input array. By changing the constant k, different networks can be developed for finding various order statistics. By a meaningful combination, any member of the OSF family can be implemented. The value of the constant k in the selection networks can also be changed adaptively. By doing so, an adaptive OSNet can be obtained which can be used to implement adaptive order statistic filters. The construction of OSNet is shown, and some examples of using OSNet to implement order statistic filters are considered


a place of mind, The University of British Columbia

Electrical and Computer Engineering
2332 Main Mall
Vancouver, BC Canada V6T 1Z4
Tel +1.604.822.2872
Fax +1.604.822.5949

Emergency Procedures | Accessibility | Contact UBC | © Copyright 2021 The University of British Columbia