The Effect of PCA On Kernel Classifiers For the Diagnosis of Alzheimer’s Disease
0
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
This paper proposes the effect of principal component analysis (PCA) on many kernel classification techniques, such as (i) Single kernel Multi-Class support vector machine (SVM), (ii) Multiple Kernel Learning (MKL) and (iii) Generalized Multiple Kernel Learning (GMKL). PCA is a sophisticated way to reduce the dimensionality of high dimensional data, specially the high dimensional imaging data, while keeping most of the informative representative dimensions and excluding the least important ones. The aim of this Paper is to figure out whether PCA is actually a good idea in practice with kernel classifiers or not, in order to find a better framework for classification and analysis of brain imaging features in the study of Alzheimer’s disease. For this experiment the data used is FreeSurfer mean cortical thickness dataset downloaded from the ADNI database (adni.loni.usc.edu) baseline data. The classification accuracy (in classifying the three classes CN, LMCI, AD) of comparative methods has been evaluated using 3-fold cross validation. As a result of this comparative study, we could observe that applying PCA on different kernel classification methods has improved the classification accuracy.
Abstract
This paper proposes the effect of principal component analysis (PCA) on many kernel classification techniques, such as (i) Single kernel Multi-Class support vector machine (SVM), (ii) Multiple Kernel Learning (MKL) and (iii) Generalized Multiple Kernel Learning (GMKL). PCA is a sophisticated way to reduce the dimensionality of high dimensional data, specially the high dimensional imaging data, while keeping most of the informative representative dimensions and excluding the least important ones. The aim of this Paper is to figure out whether PCA is actually a good idea in practice with kernel classifiers or not, in order to find a better framework for classification and analysis of brain imaging features in the study of Alzheimer’s disease. For this experiment the data used is FreeSurfer mean cortical thickness dataset downloaded from the ADNI database (adni.loni.usc.edu) baseline data. The classification accuracy (in classifying the three classes CN, LMCI, AD) of comparative methods has been evaluated using 3-fold cross validation. As a result of this comparative study, we could observe that applying PCA on different kernel classification methods has improved the classification accuracy.
참고문헌 (10)
- L. E. Hebert, "Annual Incidence of Alzheimer Disease in the United States Projected to the Years 2000 Through 2050", Alzheimer Disease and Associated Disorders, SE‐15(4), pp.169~173 (2001)
- V. Vapnik, "The Nature of Statistical Learning Theory" (1995), Springer
- C. Cortes, "Support vector networks", Machine Learning, SE‐20, pp.125~ (1995)
- B. E. Boser, "A training algorithm for optimal margin classifier", Proc. 5th ACM Workshop on Computational Learning Theory, pp.144~152 (1992)
- M. Pontil, "Support vector machines for 3-d object recognition", IEEE Trans. on Pattern Analysis and Machine Intelligence, SE‐20 (1998)
- V. Blanz, "Comparison of View-based Object Recognition Algorithms using Realistic 3D Models", Artificial Neural Networks—ICANN’96, pp.251~256 (1996)
- G. R. G. Lanckriet, "Kernel-based Data Fusion and Its Application to Protein Function Prediction in Yeast", Pacific symposium on Biocomputing (2004)
- A. Rakotomamonjy, "SimpleMKL. Journal of Machine Learning Research", Journal of Machine Learning Research, (9), pp.2491~2521 (2008)
- M.Varma, "More generality in efficient multiple kernel learning", Proc. of the International Conference on Machine Learning, pp.1065~1072 (2009)
- A. Mallak, "Effect of PCA on SVM classfier and k-means Clustering for Alzheimer's Disease Diagnosis", Proc. of 2015KING Fall Conference, pp.68~71 (2015)