頁籤選單縮合
題 名 | A Fast Learnign Algorithm of Adaptive Weighted Order Statistics Filters |
---|---|
作 者 | 姚志佳; | 書刊名 | 南開學報 |
卷 期 | 7(上) 民91.06 |
頁 次 | 頁255+257-264 |
分類號 | 440.11 |
關鍵詞 | WOS; Dichtotmy; Orthant constraint; |
語 文 | 英文(English) |
英文摘要 | In this paper, a fast and efficient learning algorithm is proposed to improve the learning speed and design capability of the adaptive weighted order statistic (WOS) filters. In this algorithm, the concept of dichotomy are adopted to reduce the learning data from each level of threshold decomposition to two levels. Moreover, Boolean function with threshold logic can be transformed into nonlinear equations with orthant constraint, and solved by two-metric projection methods. Consequently, our method can approximate the optimal weighted order statistic filters rapidly in contrast to the adaptive neural filters whose learning structure is based on the architecture of threshold decomposition. |
本系統中英文摘要資訊取自各篇刊載內容。