A robust support vector data description classifier

摘要

The conventional SVDD model is an effective tool for describing a set of training data by using proper kernel functions. Nonetheless, the SVDD could sometimes generate such a loose decision boundary while some noisy samples (outliers) exist in the training set. To solve the difficulty, we introduced a weighting to each data point in training data. The weighting can be used to measure the degree of the data point to be an outlier. By using the weighting, we reformulated a robust SVDD classifier. Experiments with various data sets showed promising results.