查詢結果分析
來源資料
相關文獻
- An On-Line Learning Neural Network for Radar Signal Classification
- 非監督類神經網路之迭代合併式學習法則
- Neural Network Procedures for Taguchi's Dynamic Problems
- A Fast and Efficient Competitive Learning Design Algorithm Based on Weight Vector Training in Transform Domain
- 專家系統振動訊號圖型判別之研究
- 反傳遞模糊類神經網路於流量推估之應用
- 類神經網路(Neural Networks)的種類及其在影像處理上的應用
- C++Fuzzy類神經網路物件導向發展系統之建立
- 臺灣汽保費率之估計--對數線性費率模式與類神經網路之比較
- 運用類神經網路於股價指數之套利--以日經225指數為例
頁籤選單縮合
| 題 名 | An On-Line Learning Neural Network for Radar Signal Classification=以線上學習之類神經網路進行雷達信號分類之研究 |
|---|---|
| 作 者 | 徐學群; 張志吉; 李岳軍; 鄧衍祥; | 書刊名 | 中正嶺學報 |
| 卷 期 | 25:2 1997.01[民86.01] |
| 頁 次 | 頁115-133 |
| 分類號 | 448.81 |
| 關鍵詞 | 非監督式學習法則; 解交錯; 類神經網路; 幅射基網路; Unsupervised learning; Deinterleaving; Neural network; Radial basis network; |
| 語 文 | 英文(English) |
| 中文摘要 | 本文提出適用於將雷達脈波解交錯(deinterleaving)的類神經網路結構與學習法 則。此非監督式學習法則包含兩個學習階段 (learning phases),我們分別稱之為聚類階段 (clustering phase) 與提煉階段 (refinement phase)。在聚類階段中,應用非監督式學習 法則於輸入層及隱藏層間以便將所接收之脈波進行分類工作。而在提煉, 階段則是使用事先 定義之價值函數 (cost function) 進行最佳化的迭代 (iterative) 學習,以調整各聚類中 心的位置。經調整後,若兩個聚類中心彼此靠得很近,則此兩個聚類將被合併。本文提出的 方法可改進非監督式學習法則(例如 ART )中分類結果與樣本學習的順序有密切關係的重 大缺點。 |
| 英文摘要 | In this paper a structure of neural network as well as a learning algorithm suitable for deinterleaving radar pulses are proposed. An unsupervised learning method which consists of two learning phases, namely, the clustering phase and the refinement phase, is introduced. In the clustering phase, the received pulses are categorized in the input layer and hidden layer of the network. In the refinement phase, a cost function is defined by the distance between members and the center for each cluster. Then the center of cluster is refined through an iterative cost optimization process. After refinement, if two cluster centers are found to be too close, they are asked to be merged into one. By the proposed method, the defect of the unsupervised network like ART where the result of classification closely depends on the learning sequence of patterns can be improved and the number of categorizing error can be reduced. |
本系統中英文摘要資訊取自各篇刊載內容。