Ranked List Loss for Deep Metric Learning

• Main Authors: Wang, Xinshao, Hua, Yang, Kodirov, Elyor, Hu, Guosheng, Garnier, Romain, Robertson, Neil M.
• Format: Other Non-Article Part of Journal/Newspaper
• Language: English
• Published: 2020
• Subjects: The Gallery; Triplets; DML
• Online Access: https://pure.qub.ac.uk/en/publications/ranked-list-loss-for-deep-metric-learning(1efa7f64-9d44-4619-940b-6e3ac4d4abed).html; https://pureadmin.qub.ac.uk/ws/files/168412256/RankedNoise.pdf

The objective of deep metric learning (DML) is to learn embeddings that can capture semantic similarity information among data points. Existing pairwise or tripletwise loss functions used in DML are known to suffer from slow convergence due to a large proportion of trivial pairs or triplets as the m...