Deep Meta Metric Learning

In this paper, we present a deep meta metric learning (DMML) approach for visual recognition. Unlike most existing deep metric learning methods formulating the learning process by an overall objective, our DMML formulates the metric learning in a meta way, and proves that softmax and triplet loss are consistent in the meta space. Specifically, we sample some subsets from the original training set and learn metrics across different subsets. In each sampled sub-task, we split the training data into a support set as well as a query set, and learn the set-based distance, instead of sample-based one, to verify the query cell from multiple support cells. In addition, we introduce hard sample mining for set-based distance to encourage the intra-class compactness. Experimental results on three visual recognition applications including person re-identification, vehicle re-identification and face verification show that the proposed DMML method outperforms most existing approaches.

1. Introduction

Distance metric learning has been widely used in many visual analysis applications, which aims to learn an embedding space where the distance between similar samples is closer and that of dissimilar samples is farther. Conventional metric learning approaches learn the embedding space by a linear Mahalanobis distance metric [13], [25], [57]. As linear metric learning approaches usually suffer from nonlinear correlations of samples, deep metric learning methods have been proposed to learn discriminative nonlinear embeddings by deep neural networks [36], [44], [55].