The kernel difference-weighted k-nearest neighbor classifier (KDF-WKNN) was superior to the nearest neighbor classifier and competitive with state-of-the-art methods for cardiac arrhythmia diagnosis.
In this paper, we proposed a kernel difference-weighted k-nearest neighbor classifier (KDF-WKNN) for the diagnosis of cardiac arrhythmia based on the standard 12 lead ECG recordings. Different from classical KNN, KDF-WKNN defines the weighted KNN rule as the constrained least-squares optimization of sample reconstruction from its neighborhood, and then uses the Lagrangian multiplier method to compute the weights of different nearest neighbors by introducing the kernel Gram matrix G. In arrhythmia analysis, it is unavoidable that some attribute values of a person would be missing. Thus, this paper further introduces a modified PCA method to address this problem. To evaluate the performance of KDF-WKNN, Experimental results on the UCI cardiac arrhythmia database indicate that, KDFWKNN is superior to the nearest neighbor classifier, and is very competitive while compared with several state-of-the-art methods in terms of classification accuracy.
Zuo et al. (Mon,) conducted a other in Cardiac arrhythmia. Kernel difference-weighted k-nearest neighbor classifier (KDF-WKNN) vs. Nearest neighbor classifier and state-of-the-art methods was evaluated on Classification accuracy. The kernel difference-weighted k-nearest neighbor classifier (KDF-WKNN) was superior to the nearest neighbor classifier and competitive with state-of-the-art methods for cardiac arrhythmia diagnosis.