Sample Selection Through Class Probability and Possibility in Support Vector Machines
Support vector machines (SVM) have shown attractive potential in classification. However, they have the limitation of size in training large data set and also sensitivity to outliers. In this paper, we used the class membership using the probability of each sample through K-nearest neighbors to reduce the training set. In SVM, if the training set contains outliers, support vectors might not be properly chosen and degrade the classification performance for unseen samples. In this case, the class membership assignment using probability is not appropriate to effectively eliminate outliers. To overcome this problem, we could check the possibility of the sample belonging to the class and eliminate the samples having weak possibility.
S. Sohn and C. H. Dagli, "Sample Selection Through Class Probability and Possibility in Support Vector Machines," Intelligent Engineering Systems Through Artificial Neural Networks, American Society of Mechanical Engineers (ASME), Jan 2001.
Engineering Management and Systems Engineering
Keywords and Phrases
Support Vector Machines (SVM); Vapnik-Chervonenkis (VC)
Article - Conference proceedings
© 2001 American Society of Mechanical Engineers (ASME), All rights reserved.