Background: It presents a kind of multi-class image classification algorithm which is
combined with Best Versus Second Best (BVSB) active learning technology and improved self-training
semi-supervised learning technology.
Methods: The algorithm integrates the advantages of active learning; semi-supervised learning and
extreme learning machine simultaneously. It has better performance than that of single algorithm when
it is used in different sets of image target recognition.
Results: In addition, it also discussed the influence of various parameters on the algorithm performance
in the experimental parts, and made related analysis of semi-supervised learning algorithm based on
SVM (Support Vector Machine); the experimental results verified the superiority of proposed