度量学习的任务是使用算法来学习输入对象之间的距离。它可以用于人脸识别认证，商品检索，通过训练的模型可以让相似对象之间的特征距离相近，不相似的对象之间的特征距离较远。深度学习模型能够有效学习这样的特征。本篇文章会介绍深度度量学习的基本原理和一些最近的进展。

什么是度量学习？

深度度量学习

  • Contrastive Loss

R. Hadsell, S. Chopra, and Y. LeCun. Dimensionality reduction by learning an invariant mapping

  • Semi-Hard Mining Strategy

F. Schroff, D. Kalenichenko, and J. Philbin. Facenet: A unified embedding for face recognition and clustering. In CVPR, 2015

  • Lifted Structure Loss

Deep Metric Learning via Lifted Structured Feature Embedding

  • Binomial BinDeviance lLoss

D. Yi, Z. Lei, and S. Z. Li. Deep metric learning for practical person re-identification

  • NCA Loss

C. Wu, R. Manmatha, A. J. Smola, and P. Kr¨ahenb¨uhl. Sampling matters in deep embedding learning. ICCV, 2017

  • Proxy-NCA

No Fuss Distance Metric Learning using Proxies

  • N-pair loss

Improved Deep Metric Learning with Multi-class N-pair Loss Objective

  • Clustering loss

Deep Metric Learning via Facility Location

  • Angular loss

Deep Metric Learning with Angular Loss

  • Multi-Similarity Loss

https://github.com/MalongTech/research-ms-loss

  • Proxy Anchor Loss

参考