Quantized Kernel Learning Filter with Maximum Mixture Correntropy Criterion

Kernel recursive maximum mixture correntropy criterion (KRMMCC) algorithm takes the mixture of two Gaussian kernel functions as the core function, which further improves the performance of machine learning. However, when the number of training data is large, the KRMMCC algorithm will face a large amount of computation. In order to restrain the growth of the radial basis function structure, this paper proposes a novel method named quantized kernel recursive maximum mixture correntropy criterion (QKRMMCC). This method judges whether the data is quantized to the nearest node by quantization rule, so as to sparse the final network size. The simulation results show that the QKRMMCC algorithm presents the excellent performance.