In this paper, a simple order filtering neural network, which can progressively output the value of a specific ordered input, is proposed. The progressive order filtering neural network (POFNN) is developed in a simple two-layer structure embedded with robustness. Both theoretical analyses and simulated results show that the designed order filtering neural network converges to the solution in logarithmic function of the dynamic range and the accuracy threshold, which are independent of the number of inputs.
|Number of pages||7|
|Journal||Journal of the Chinese Institute of Engineers, Transactions of the Chinese Institute of Engineers,Series A/Chung-kuo Kung Ch'eng Hsuch K'an|
|Publication status||Published - 1996 Jan 1|
All Science Journal Classification (ASJC) codes