支撐矢量預選取的雙色Voronoi圖方法
Pre-extracting support vector for support vector maching using bi-color voronoi diagrams
-
摘要: 支撐矢量機是在統(tǒng)計學習理論的基礎上發(fā)展出來的一種新的模式識別方法，在解決小樣本、非線性及高維模式識別問題中表現(xiàn)出許多特有的優(yōu)勢，在支撐矢量機中，支撐矢量的選取相當困難,成為其應用的瓶頸問題。該文利用Voronoi圖在特征空間特有的構造特性，提出了一種預先選取支撐矢量的新方法雙色Voronoi圖方法。該方法針對數(shù)據(jù)在空間的分布特性，在訓練支撐矢量機以前，利用樣本數(shù)據(jù)的雙色Voronoi圖確定候選的支撐矢量，然后在這些預選的矢量上進行學習。試驗證明了該方法的有效性及可行性。Abstract: Support Vector Machines (SVMs) are a new generation learning system based on recent advances in statistical learning theory. SVMs have many well features that make them attractive for small samples, nonlinear and high dimensional pattern recognition. However, choice of Support Vectors(SVs) is difficult in SVMs, which is a bottleneck problem. In this paper, a novel method using bi-color Voronoi diagram is proposed to pre-extract SVs based on Voronoi diagram. Considering the distribution feature of samples space, this method determi-nates SVs based on the bi-color Voronoi diagram before training SVMs. Learning is based on these pre-extracted vectors. Experiments show that this method is feasible and effective.
-
V.N.Vapnik著,張學工譯,統(tǒng)計學習理論的本質[M].北京,清華大學出版社,2000年,11-126.[2]V.N.Vapnik,An overview of statistical learning theory,IEEE Trans.on Neural networks,1999,10(5),988-999.[3]邊肇祺,張學工,模式識別[M],北京,清華大學出版社,2000年,161-176,284-305.[4]焦李成,張莉,周偉達,支撐矢量預選取的中心距離比值法,電子學報,2001,29(3),383-386.[5]周培德,計算幾何--算法分析與設計[M],北京,清華大學出版社,2000年,88-130,236-271.[6]F. Aurenhammer, Voronoi diagrams-a survey of a fundamental geometric data structure, ACM Comput. Survey, 1991, 23(3), 345-405.[7]G.W. Rogers, J. Solka, D. S. Malyevac, C. E. Priebe, A self-organizing network for computing a posteriori conditional class probability, IEEE Trans. on Systems, Man and Cybernetics, 1993,23(6), 1672-1682.[8]N.K. Bose, A. K. Garga, Neural network design using Voronoi diagrams, IEEE Trans. on Neural Networks, 1993, 4(5), 778-787.[9]C. Gentile, M. Sznaier, An improved Voronoi-diagram-based neural net for pattern classification,IEEE Trans. on Neural Networks, 2001, 12(5), 1227-1234.[10]閻平凡,張長水,人工神經(jīng)網(wǎng)絡與模擬進化計算[M],北京,清華大學出版社,2000年,60-95.[11]D.Chen,Efficient geometric algorithm on the EREW PRAM,IEEE Trans.on Parallel Distrib.Syst.,1995,6(1),41-47. -
計量
- 文章訪問數(shù): 2241
- HTML全文瀏覽量: 105
- PDF下載量: 544
- 被引次數(shù): 0
下載:
