Learning discriminative topological structure information representation for 2D shape and social network classification via persistent homology

Changshuo Wang, Rongsheng Cao, Ruiping Wang

Abstract: Highlights•We proposed the novel TSNet for topological feature extraction.•TS-Layer transforms persistence diagrams to vectors.•RA-Layer enhance feature discrimination.•Tests show 1.7% and 2.1% accuracy gains in 2D shape and network classification.