類神經網路之知識挖掘

Abstract

類神經網路已廣為人們接受並應用於各種不同的領域，但由於其訓練所獲得的結果，無法以一簡單的數學模式表示，使得類神經網路模式無法擁有一適當的解釋能力。目前大多數的學者使用規則萃取的方式發展各種演算法，首先簡化網路的結構，再由簡化的網路中萃取規則；而目前所發展出的演算法當中，大多僅僅是藉由網路中多餘連結之刪除來進行網路結構之簡化，且對輸入的資料型態也有所限制。本研究除了提出網路修剪及規則萃取的演算法外，也提出關於重要輸入節點選取的演算法。本研究所提出的演算法對於輸入的資料型態並沒有限制，資料型態可為二元、離散或連續，甚至混合的資料型態均可使用。最後，本研究利用二個案例，藉由比較本研究所提方法與See5分析結果，顯示本研究所提方法的有效性。

Despite their diverse applications in many domains, neural networks are difficult to interpret owing the lack of a mathematical model to express its training result. While adopting the rule extraction method to develop different algorithms, many researchers normally simplify a network's structure and then extract rules from the simplified networks. Such conventional approaches are subject to the type of data while attempting to remove unnecessary connections. In addition to developing network pruning and extraction algorithms, this work attempts to determine the important input nodes. In the proposed algorithms, the type of input data is not limited to binary, discrete or continuous. Moreover, two numerical examples are analyzed. Comparing the results of the proposed algorithms with those of See5 demonstrates the effectiveness of the proposed algorithms.