Social Learning under Uncertainty Based on Dempster-Shafer Approach for Minimizing True Error of MachineLearning

摘要

Minimizing true error of the classification process under uncertainty is one of the difficult issues in the field of machine learning. Researchers do not address this topic until this time despite its importance in practical life. This paper can be considered as a development of the concept of social learning presented the intellectual leap in the machine learning area as given before by the authors. Novelty in this paper is to present a new approach that can deal with the conditions of uncertainty resulting from multiple sources. This paper also presents a new method of social learning based on benefits offered by the Dempster-Shafer theory (DST) of evidence. The paper provides experimental results on six benchmarks. The results attained from the comparison using six benchmarking problems illustrate a superior performance of the proposed method compared with the best results attained in the literature of machine learning domain till now.