DiscFace: Minimum Discrepancy Learning for Deep Face Recognition
Insoo Kim (Samsung Advanced Institute of Technology)*, Seungju Han (Samsung Advanced Institute of Technology), Seong-Jin Park (Samsung Advanced Institute of Technology), Ji-won Baek (Samsung Advanced Institute of Technology), Jinwoo Shin (KAIST), Jae-Joon Han (Samsung), Changkyu Choi (Samsung)
Keywords: Face, Pose, Action, and Gesture
Abstract:
Softmax-based learning methods have shown state-of-the-art performances on large-scale face recognition tasks. In this paper, we discover an important issue of softmax-based approaches: the sample features around the corresponding class weight are similarly penalized in the training phase even though their directions are different from each other. This directional discrepancy, i.e., process discrepancy leads to performance degradation at the evaluation phase. To mitigate the issue, we propose a novel training scheme, called minimum discrepancy learning that enforces directions of intra-class sample features to be aligned toward an optimal direction by using a single learnable basis. Furthermore, the single learnable basis facilitates disentangling the so-called class-invariant vectors from sample features, such that they are effective to train under class-imbalanced datasets.
SlidesLive
Similar Papers
Exploiting Transferable Knowledge for Fairness-aware Image Classification
sunhee hwang (Yonsei university)*, Sungho Park (Yonsei University), Pilhyeon Lee (Yonsei University), seogkyu jeon (Yonsei university), Dohyung Kim (Yonsei University), Hyeran Byun (Yonsei University)

Imbalance Robust Softmax for Deep Embedding Learning
Hao Zhu (Australian National University)*, Yang Yuan (AnyVision), Guosheng Hu (AnyVision), Xiang Wu (Reconova), Neil Robertson (Queen's University Belfast)

Few-Shot Zero-Shot Learning: Knowledge Transfer with Less Supervision
Nanyi Fei (Renmin University of China), Jiechao Guan (Renmin University of China), Zhiwu Lu (Renmin University of China)*, Yizhao Gao (Renmin University of China)
