Masked contrastive graph representation learning for age estimation

Yuntao Shou, Xiangyong Cao, Huan Liu, Deyu Meng

Abstract: Highlights•The Proposed MCGRL method alleviates the inflexibility in modeling irregular objects.•The MCGRL method uses a masked GCN to improve the features representation. ability.•The complementary semantic information is captured by contrastive learning.•The proposed method outperformed SOTA methods on three public datasets.