Abstract:One of the main challenges for feature representation in deep learning-based classification is the design of appropriate loss functions that exhibit strong discriminative power. The classical softmax loss does not explicitly encourage discriminative learning of features. A popular direction of research is to incorporate margins in well-established losses in order to enforce extra intra-class compactness and inter-class separability, which, however, were developed through heuristic means, as opposed to rigorous mathematical principles. In this work, we attempt to address this limitation by formulating the principled optimization objective as learning towards the largest margins. Specifically, we firstly propose to employ the class margin as the measure of inter-class separability, and the sample margin as the measure of intra-class compactness. Accordingly, to encourage discriminative representation of features, the loss function should promote the largest possible margins for both classes and samples. Furthermore, we derive a generalized margin softmax loss to draw general conclusions for the existing margin-based losses. Not only does this principled framework offer new perspectives to understand and interpret existing margin-based losses, but it also provides new insights that can guide the design of new tools, including \textit{sample margin regularization} and \textit{largest margin softmax loss} for class balanced cases, and \textit{zero centroid regularization} for class imbalanced cases. Experimental results demonstrate the effectiveness of our strategy for multiple tasks including visual classification, imbalanced classification, person re-identification, and face verification.

Large margin nearest local mean classifier

Hierarchical Large Margin Nearest Neighbor Classification

Hierarchical Distance Metric Learning for Large Margin Nearest Neighbor Classification.

A Multi-Class Large Margin Classifier

Efficiently Learning a Distance Metric for Large Margin Nearest Neighbor Classification.

CLE_LMNN

Scalable Large-Margin Mahalanobis Distance Metric Learning

Local Mean Representation Based Classifier and Its Applications for Data Classification.

Improving the discriminant ability of local margin based learning method by incorporating the global between-class separability criterion

Large-margin Nearest Neighbor Classifiers Via Sample Weight Learning.

Parameter Free Large Margin Nearest Neighbor For Distance Metric Learning

Relative Local Mean Classifier with Optimized Decision Rule

Learning Towards The Largest Margins.

Unified Locally Linear Classifiers with Diversity-Promoting Anchor Points

Large Margin Discriminative Loss for Classification

LAM3L: Locally Adaptive Maximum Margin Metric Learning for Visual Data Classification

Two Dimensional Large Margin Nearest Neighbor for Matrix Classification.

Local and Weighted Maximum Margin Discriminant Analysis

Regularized Large Margin Distance Metric Learning.

A Large Margin Classifier with Additional Features

Large Margin Weighted K-Nearest Neighbors Label Distribution Learning for Classification.