GCIM based Improved Proportionate APA for Cluster-Sparse Identifications
Biao Xie, Shengwei Xie, Ji Zhao, Qiang Li · 2024
In cluster-sparse system identification, a series of algorithms have been proposed by solving the cost function with a cluster-sparse penalty. Inserting the mixed the cluster-sparse improved proportionate affine projection algorithm (CS-IPAPA) has been derived for cluster-sparse system. However, for CS-IPAPA, the mixed may not be the best choice compared to the mixed Therefore, injecting the and generalized correntropy induced metric (GCIM) into the mixed problems of CS-IPAPA, we propose a family of GCIM-based CS-IPAPA from a basis pursuit perspective. In addition, considering the computational cost of exponential operations and the historical information of the proportional matrix, we derive a simplified version of CS-GCIM-IPAPA with the Taylor expansion method and time-shift property. Finally, compared to some competing sparse algorithms, we conduct a series of simulations to verify the superior filtering performance of our proposed algorithms with different input signals and various parameters.