Performance Validation of Jointly Optimized Graph Filters

In this paper, a novel jointly optimized graph filter is investigated, which is based on the vertex sampling method and the sum-of-squares integration method. It takes the frequency response of the ideal graph filter and the error value of the frequency response of the designed graph filter as the optimization objectives of the jointly optimized graph filter. Firstly, the theoretical analysis and validation of the practical performance of this method are carried out in this paper, and theoretically, we can get that the performance of the jointly optimized graph filter is improved compared with that of the non-jointly optimized graph filter. Then, the actual performance of the jointly optimized graph filter is verified using real data sets. Finally, the jointly optimized graph filter design method is compared with two benchmark graph filter design methods. It is concluded that the performance of the jointly optimized graph filter also has a good improvement over the benchmark graph filters.