Abstract: Automatic identification of insects is an important and emerging area of research. The screening features and transforming multi-class classification into two-class classification properly are two key procedures in the process. In this article, a novel method for automatic identification of multi-class insects was developed based on support vector classification (SVC). Firstly, the initial multi-class samples were transformed into two-class samples with interaction transformation. Secondly, a symmetrical kernel function was inducted to solve the rank problem of the two initial samples in interaction sampling pair. Thirdly, irrelevant and redundant features were eliminated nonlinearly with SVC and the relative importances of kept features were listed. Lastly, the prediction results were further corrected by simple-vote decision. The new method was applied to identify the butterflies of seven species at species level and family level, and the accuracies at both levels are 100%. The results show that the new method can be widely used in the prediction area of multi-class classification, such as automatic identification of insects.

Key words: Support vector classification (SVC), butterfly, automatic identification, screening features, interaction transformation