Natural neighborhood-based classification algorithm without parameter k

Abstract

Various kinds of k-Nearest Neighbor (KNN) based classification methods are the bases of many wellestablished and high-performance pattern recognition techniques. However, such methods are vulnerable to parameter choice. Essentially, the challenge is to detect the neighborhood of various datasets while ignoring the data characteristics. This article introduces a new supervised classification algorithm, Natural Neighborhood Based Classification Algorithm (NNBCA). Findings indicate that this new algorithm provides a good classification result without artificially selecting the neighborhood parameter. Unlike the original KNN-based method, which needs a prior k, NNBCA predicts different k for different samples. Therefore, NNBCA is able to learn more from flexible neighbor information both in the training and testing stages. Thus, NNBCA provides a better classification result than other methods.