遞增式模糊分類系統

An Incremental Fuzzy Rule-based Classification System

傅仕熙(Shih-Hsi Fu)

指導教授 ： 阮議聰

中原大學/工學院/資訊工程研究所/碩士(2003年)

https://doi.org/10.6840/cycu200300324

摘要

在本論文中我們提出一個遞增式模糊分類系統，主要目的是為了產生一個線上的、動態的分類系統，能隨著資料的增加，不須重頭建構，系統更新是遞增的，不忘卻先前的知識，建構時間較快速。使用簡單模糊格狀方法(Simple Fuzzy Grid Method)及CART (Classification and Regression Tree)的資訊增益(information gain)從已知類別的數值資料來自動建構遞增式模糊分類系統。簡單模糊格狀方法是用來建構初始的模糊分類系統，由於此方法產生的分類系統，在訓練後不需要學習，就有基本分類能力。當有額外的資料進來時，利用每個模糊規則的確定性因子(certainty factor)來判定模糊規則是否需要再分割，而CART的資訊增益是用來決定模糊規則如何再分割。
為了減少模糊規則個數，使得每個模糊規則更一般性，減少資料無法推論出類別的情形發生，因此我們考慮模糊規則可以與鄰近的模糊規則合併，如此不但能夠增加分類能力，也能減少模糊規則個數。最後再利用Iris及glass資料庫來測試遞增式模糊分類系統在合併前與合併後的學習能力、推廣能力及模糊規則個數，並與其他模糊分類系統比較。

關鍵字：模糊分類系統、遞增式學習、決策樹、資訊理論

關鍵字

模糊分類系統 ； 遞增式學習 ； 資訊理論 ； 決策樹

並列摘要

In this paper, we propose an incremental fuzzy rule-based classification system. The main objective is to generate an on-line, dynamic, incremental classification system. A method for automatic construction of an incremental fuzzy rule-based classification system from numerical data using the Simple Fuzzy Grid Method and the information gain of the Classification and Regression Tree (CART) is presented. The Simple Fuzzy Grid Method is developed for an initial fuzzy rule-based classification system from training patterns. When incremental data input, exploiting certainty factor of each fuzzy if-then rule determines whether fuzzy if-then rule need to be repartitioned or not. And the information gain of CART decides how to repartition fuzzy if-then rule.
In order to reduce the number of fuzzy if-then rules and to make each fuzzy if-then rule more general, we consider combining two adjacent fuzzy if-then rules. Finally we use Iris and glass database to estimate learning ability, generalization ability and the number of fuzzy if-then rules of the incremental fuzzy rule-based classification system before and after combining. And we compare above characteristics with other fuzzy classification systems.

Keyword：Fuzzy Rule-based Classification System, Incremental Learning, Decision Tree, Information Theory

並列關鍵字

Incremental Learning ； Decision Tree ； Fuzzy Rule-based Classification System ； Information Theory

參考文獻

[1] H. Ishibuchi, T. Nakashima, and T. Murata, “ Performance evaluation of fuzzy classifier systems for multi-dimensional pattern classification problems ,” IEEE Trans. Syst., Man, Cybern. B, vol. 29, pp. 601–618, Oct. 1999.

[2] Ludmila I. Kuncheva, “How good are fuzzy if-then classifiers,” IEEE Trans. Syst., Man,Cybern. B, vol. 30, pp. 501–509, Aug. 2000.

[3] H. Ishibuchi, T. Nakashima, “Effect of Rule Weights in Fuzzy Rule-Based Classification Systems,” IEEE Trans. Fuzzy Syst., vol. 9, no. 4, pp. 506–515, Aug. 2001.

[4] Oscar Cordón, Francisco Herrera, and Pedro Villar, “Generating the Knowledge Base of a Fuzzy Rule-Based System by the Genetic Learning of the Data Base,” IEEE Trans. Fuzzy Syst., vol. 9, no. 4, pp. 667–674, Aug. 2001.

[5] O. Cordón, M.J. del Jesus, F. Herrera, “Genetic Learning of Fuzzy Rule-Based Classification Systems Cooperating with Fuzzy Reasoning Methods,” International Journal of Intelligent Systems 13, pp. 1025-1053, 1998