A fast learning algorithm for One-Class Slab Support Vector Machines

摘要

One-Class Slab Support Vector Machines (OCSSVM) have turned out to be better in terms of accuracy in certain types of classification problems than the traditional SVMs, One Class SVMs, and other one-class classifiers. This paper proposes a fast training method for the OCSSVM using a modified Sequential Minimal Optimization (SMO) algorithm, which would enhance its scalability without a significant compromise in precision. We compared our SMO-based algorithm, the regular OCSSVM, and other state-of-the-art one-class classifiers for time and accuracy on multiple benchmark datasets. The experimental results indicate that the proposed training method provides the best tradeoff between training time and accuracy among the compared methods. It achieves accuracies similar to the regular OCSSVM and better or comparable to existing state-of-the-art one-class classifiers. It provides better scalability than the regular OCSSVM and most other classifiers.