Multiple birth support vector machine based on recurrent neural networks

摘要

Multiple birth support vector machine (MBSVM) is a new classification algorithm, which includes the advantages of low complexity and high computing efficiency. However, the traditional MBSVM does not take into account the correlation sequence information among all dimensions of the samples when using the method to classify datasets, which limits the further improvement of the classification accuracy. Although some scholars have combined neural networks with support vector machine (SVM), these methods do not take into account the sequence correlation among different features. For the above problems, we present several variants of MBSVM algorithms to illustrate the validity and reliability of the theory: Multiple Birth Support Vector Machine based on Multilayer Perceptron (MLP-MBSVM), Multiple Birth Support Vector Machine based on Long-Short Term Memory Networks (LSTM-MBSVM), Multiple Birth Support Vector Machine based on Multilayer Perceptron and Long-Short Term Memory Networks(MLP-LSTM-MBSVM). After introducing multilayer perceptron and long-short term memory networks, these algorithms can take full account of the sequence correlation information between different features of samples. The experiments results show that the algorithms proposed in this paper are effective, and they can greatly improve the classification accuracy of multiple birth support vector machine.