Semi-Supervised Representation Learning via Triplet Loss Based on Explicit Class Ratio of Unlabeled Data

Kazuhiko Murasaki, Shingo Ando, Jun Shimamura