Abstract:The objective of deep metric learning (DML) is to learn embeddings that can capture semantic similarity and dissimilarity 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 model improves. To improve this, ranking-motivated structured losses are proposed recently to incorporate multiple examples and exploit the structured information among them. They converge faster and achieve state-of-the-art performance. In this work, we unveil two limitations of existing ranking-motivated structured losses and propose a novel ranked list loss to solve both of them. First, given a query, only a fraction of data points is incorporated to build the similarity structure. Consequently, some useful examples are ignored and the structure is less informative. To address this, we propose to build a set-based similarity structure by exploiting all instances in the gallery. The learning setting can be interpreted as few-shot retrieval: given a mini-batch, every example is iteratively used as a query, and the rest ones compose the gallery to search, i.e., the support set in few-shot setting. The rest examples are split into a positive set and a negative set. For every mini-batch, the learning objective of ranked list loss is to make the query closer to the positive set than to the negative set by a margin. Second, previous methods aim to pull positive pairs as close as possible in the embedding space. As a result, the intraclass data distribution tends to be extremely compressed. In contrast, we propose to learn a hypersphere for each class in order to preserve useful similarity structure inside it, which functions as regularisation. Extensive experiments demonstrate the superiority of our proposal by comparing with the state-of-the-art methods on the fine-grained image retrieval task. Our source code is available online: https://github.com/XinshaoAmosWang/Ranked-List-Loss-for-DML.

Deep Metric Learning with Tuplet Margin Loss.

Deep Metric Learning with Dynamic Margin Hard Sampling Loss

Learning Deep Embeddings Via Margin-Based Discriminate Loss.

Adaptive Few-Shot Deep Metric Learning

Deep Metric Learning: Loss Functions Comparison

Deep Metric Learning with Dynamic Margin Hard Sampling Loss for Face Verification

Deep Ranking with Adaptive Margin Triplet Loss

SoftTriple Loss: Deep Metric Learning Without Triplet Sampling

Ranked List Loss for Deep Metric Learning

Learnable dynamic margin in deep metric learning

Deep Adversarial Metric Learning

Deep Metric Learning with Hierarchical Triplet Loss.

The General Pair-based Weighting Loss for Deep Metric Learning

Soft Margin Triplet-Center Loss for Multi-View 3D Shape Retrieval

Contrastive Bayesian Analysis for Deep Metric Learning

Metric Learning Via Penalized Optimization.

Margin Loss Based on Adaptive Metric for Image Recognition

Deep Metric Learning for Text Data Based on Triplet Network

Regularized Large Margin Distance Metric Learning.

Mean Field Theory in Deep Metric Learning

Concordance Learning with Spectral Decay on Triplet Similarity