A Fuzzy KNN Algorithm Based on Weighted Chi-square Distance
Xingang Wang, Peipei Yao · Proceedings of the 2nd International Conference on Computer Science and Application Engineering · 2018
The traditional k Nearest Neighbor (KNN)1 algorithm does not consider the relative relationship between the sample features. The classification speed is slow and the computational complexity is high. The distance between the test sample and all the training samples needs to be calculated to determine the k nearest neighbors. Therefore, this paper proposes a fuzzy k nearest neighbor (FKNN) algorithm based on weighted chi-square distance. First, the fuzzy normalization process is performed, and the similarity is taken as the fuzzy membership degree. The closeness of features is used to determine the weight of each feature, and the weighted chi-square distance is used as the distance measure. Finally, the sample class to be classified is determined by the class membership of k neighbors. The classification results show that the evaluation indexes of the algorithm are better than the existing ones.