Deep Metric Learning With Density Adaptivity

Abstract

The problem of distance metric learning is mostly considered from the perspective of learning an embedding space, where the distances between pairs of examples are in correspondence with a similarity metric. With the rise and success of Convolutional Neural Networks (CNN), deep metric learning (DML) involves training a network to learn a nonlinear transformation to the embedding space. Existing DML approaches often express the supervision through maximizing inter-class distance and minimizing intra-class variation. However, the results can suffer from overfitting problem, especially when the training examples of each class are embedded together tightly and the density of each class is very high. In this paper, we integrate density, i.e., the measure of data concentration in the representation, into the optimization of DML frameworks to adaptively balance inter-class similarity and intra-class variation by training the architecture in an endto-end manner. Technically, the knowledge of density is employed as a regularizer, which is pluggable to any DML architecture with different objective functions such as contrastive loss, N-pair loss and triplet loss. Extensive experiments on three public datasets consistently demonstrate clear improvements by amending three types of embedding with the density adaptivity. More remarkably, our proposal increases Recall@1 from 67.95% to 77.62%, from 52.01% to 55.64% and from 68.20% to 70.56% on Cars196, CUB-200-2011 and Stanford Online Products dataset, respectively.

Smart Director: An Event-Driven Directing System for Live Broadcasting

“…The higher the score, the more chance the view (i.e., video clip) to be selected for highlight. Inspired by the relative relation modeling via ranking [22,24,30,31,33,47,50], one way to train the MVHD is to utilize the pairwise ranking objective that enforces the score of highlighted clip higher than that of non-highlighted clip. Nevertheless, this will result in a sub-optimal solution, because the pairwise ranking objective can only compare one pair of highlighted and non-highlighted clips, leaving other non-highlighted views unexploited.…”

Deep Metric Learning With Manifold Class Variability Analysis

“…EEP metric learning (DML) has been used to compare the similarity of samples in a supervised or unsupervised manner, and has been applied to various fields such as product search [1,2], and video highlight detection [3]. A typical way of DML is to utilize the triplet loss that defines the triangular relationship between samples in terms of Euclidean distance [4,5].…”

Probabilistic Principal Geodesic Deep Metric Learning

“…Note that most DML studies only considered local similarity based on pairwise connection. Since DML must be able to improve the retrieval performance, global features such as semantic representation need to be reflected in the similarity metric [13,66]. In addition, since existing DML methods defined the embedding space produced by convolutional neural networks (CNNs) as a vector space, they had a limit in reflecting nonlinear characteristics such as multi-variate covariance (see Sec.…”