Application on small sample classification of ER rule classifier

Most classical classification methods require a large number of samples to be trained, so the classification accuracy will decline when the sample size is not large enough. Considering the parameters of the recently proposed Evidential Reasoning rule (ER rule) classifier can be set according to experience and expert knowledge, a small sample classification method based on the ER rule classifier is proposed in this study. The classification steps of ER rule classifier contain evidence acquisition, reliability calculation, aggregation of activated evidence, and optimization of the adjustable parameters. And the case study about sleep/wake classification is conducted, including experiments of classification by small sample and 10-fold cross-validation based on ER rule classifier, and classification by small sample based on six classical classification methods. Finally, the feasibility and effectiveness of the ER Rule classifier to classify small samples with high accuracy are verified by analyzing the results.