Study On Incremental Learning Algorithm Based On Mahalanobis Hyper Ellipsoidal Learning Machine

Support vector machines(SVMs), as a new machine learning method based on statistical learning theory and structural risk minimization, have attracted more and more attention and became a hot issue in the field of machine learning, because it can well resolve such practical problems as nonlinearity, high dimension and local minima, and has found a great deal of success in a wide range of applications including pattern recognition and text categorization. SVM has a good generalization performance because it does not depend on all the training data, but a subset named support vector, so SVM is suit to large data and is a powerful tool to the incremental learning. It has the important theory significance and practical value that study on SVM incremental learning algorithm.A Mahalanobis hyper ellipsoidal learning machine class incremental learning algorithm is proposed. To each class sample, the hyper ellipsoidal that enclose as much as possible and push the outlier samples away are trained in the feature space. In the process of incremental learning, only one sub-classifier is trained with the new class samples. The old models of the classifier are not influenced and can be reused. In the process of classification, considering the information of sample’s distribution in the feature space, the Mahalanobis distances from the sample mapping to the center of each hyper ellipsoidal are used to decide the classified sample class. The experimental results show that the proposed method has higher classification precision and classification speed.A Mahalanobis hyper ellipsoidal support vector machines incremental learning algorithm is proposed. To each class sample, the hyper ellipsoidal that includes as much as possible and push the outlier samples away are trained in the feature space. In the process of incremental learning, only the new incremental samples and the old class support vectors those have the same classes as the new incremental samples are trained or retrained. For the sample to be classified, the Mahalanobis distances from the sample mapping to the center of each hyper ellipsoidal are used to decide the sample class. The experimental results show that the proposed method has higher performance on training speed, classification precision and classification speed.