1: \begin{abstract}

2:   % motivation problem approach results conclusion (impact)

3: %   In deep metric learning (DML), we are interested in learning to embed images as feature vectors whose distances are useful for tasks such as image retrieval. A number of loss functions have been proposed for Deep Metric Learning (DML) in recent years, among which, proxy-based losses are superior in their fast convergence and low training complexity.

4:   Proxy-based metric learning losses are superior to pair-based losses due to their fast convergence and low training complexity. However, existing proxy-based losses focus on learning class-discriminative features while overlooking the commonalities shared across classes which are potentially useful in describing and matching samples. Moreover, they ignore the implicit hierarchy of categories in real-world datasets, where similar subordinate classes can be grouped together. In this paper, we present a framework that leverages this implicit hierarchy by imposing a hierarchical structure on the proxies and can be used with any existing proxy-based loss. This allows our model to capture both class-discriminative features and class-shared characteristics without breaking the implicit data hierarchy. We evaluate our method on five established image retrieval datasets such as In-Shop and SOP. Results demonstrate that our hierarchical proxy-based loss framework improves the performance of existing proxy-based losses, especially on large datasets which exhibit strong hierarchical structure.

5:   \blfootnote{*Work partially done while interning at Amazon.}

6: \end{abstract}

7: