Research And Implementation Of Incremental Learning Based On Support Vector Machine

The explosion of the network information to the people brought new challenges,the era of Web page classification is to organize and make use of large amounts of Internet information is an effective way.Widely used in web search engine,support vector machine(SVM)showed excellent learning and memory function,has become a hot research topic in the field of machine learning.Classic support vector machine training algorithm does not support incremental learning,if again for all the training samples is very waste of time.Therefore,the focus of future development in the IT industry will be focused on the incremental learning method of support vector machine(SVM).The paper researches and introduces the development of support vector machine,principle and techniques.It expounds support vector machine training algorithm and incremental learning ways.The paper expounds the principle and mechanism of kernel function of support vector machine,discussing the advantages and disadvantages of global kernel function and local kernel function.n the incremental learning of support vector machine,the optimal hyperplane will rotate when the new training data is added.Thus incremental learning candidates of support vector probably tend to be deleted.For gaining more precise knowledge,the paper proposes a novel screening method to select support vector candidates.It shows that hypercone model can get more precise knowledge for incremental training and cut down the consumption time.For local kernel function is good but the poor generalization ability,learning ability is put forward combining global kernel function to construct the new joint function method,to combine the advantages of the two kinds of kernel function,it better adapt to incremental learning and improve the classification accuracy.Finally,the improved training algorithm is applied to the web page classification system,and the improved algorithm has carried on the contrast experiment and performance analysis.The experimental data show that the algorithm has higher classification efficiency and accuracy.