國立臺灣大學電機工程學系暨研究所王勝德2006-07-252018-07-062006-07-252018-07-062003-07-31http://ntur.lib.ntu.edu.tw//handle/246246/7917學習演算法(Learning Algorithms)是智慧型代理人(Intelligent Agent)中最基本 的組成部分，藉由不同的學習演算法，可使智慧型代理人表現出不同的效能和行 為。在眾多學習演算法中，分類演算法可使智慧型代理人從已知資料中學習分類 的方法，進而從未來的資料中做出正確的預測或表現合適的行為。模糊向量支持 機器(Fuzzy support vector machines, FSVMs)將每一個資料點聯結一個可表示資 料意義的模糊成員函數，降低雜訊在學習過程的影響。在這篇報告中，我們提出 及比較兩種自動設定資料點模糊成員函數的方法，使得我們能更方便地處理有雜 訊的分類問題及應用模糊向量支持機器於智慧型代理人中。使用指標性的資料庫 與其他演算法做比較，實驗結果證明我們的演算法可以有效的處理這個問題。Learning algorithms are the basic part of intelligent agent. Different learning algorithms can affect the performance and behavior of the intelligent agent. Classification algorithms used in many intelligent agent applications are one of these learning algorithms. Intelligent agents can learn the behavior from known data and predict or behave from future data by classification algorithms. Fuzzy support vector machines (FSVMs) provide a way to classify data with noises or outliers. Each data point is associated with a fuzzy membership that can reflect their relative degrees as meaningful data. In this report, we investigate and compare two strategies of automatically setting the fuzzy memberships of data points. It makes the usage of FSVMs easier in the application of reducing the effects of noises or outliers such that makes better behavior of intelligent agents. The experiments show that the generalization error of FSVMs is comparable to other methods on benchmark datasets.application/pdf226133 bytesapplication/pdfzh-TW國立臺灣大學電機工程學系暨研究所分類機器學習雜訊模糊成員函數classificationmachine learningnoisefuzzy membershipreporthttp://ntur.lib.ntu.edu.tw/bitstream/246246/7917/1/912213E002044.pdf