An improved SVDU-IKPCA algorithm for Specific Emitter Identification
Dan Xu, Bo Yang, Wenli Jiang, Yiyu Zhou · 2008
A forecast learning method of kernel principal component analysis (KPCA) is presented for specific emitter identification (SEI) application. By constructing a symmetrical decomposition of the kernel matrix, we derived a new algorithm of incremental KPCA. Based on it, the forecast capability is developed by creating dummy samples whose kernel vectors are an extrapolation of the kernel matrix. The advance of the algorithm is verified in the SEI numerical experiment.