結合核心主成分與支持向量機以提昇資料分類效能

To Improve the Classification Performance by CombiningKPCA with Support Vector Machine

結合核心主成分分析（KPCA）與支持向量機（SVM）模式，先將資料透過核心函數映射至高維空間進行核主成分轉換，以獲取高階訊息的主要成分，做為資料的前置處理。然後將處理過後的資料當做支持向量機的輸入來測試經由核心主成分特徵變換是否能提昇資料分類效能。實驗是利用UCI資料庫中的二個資料集，實驗結果說明KPCA+SVM模式對經由核主成分徵變換後的資料分類，確實可以提昇分類的正確性。

The purpose of this paper is to apply kernel principal component analysis (KPCA) to support vector machine (SVM) for data classification. The first, to translate data into high dimension space via kernel function that we can get from the kernel principal components. The second, to put these components into SVM, then watch the results whether are more powerful than only SVM does. The Experiment shows the model of KCPA+SVM has more efficient by using kernel principal component analysis in data classification, from two datasets in UCI.