Virtual Class Enhanced Discriminative Embedding Learning

TL;DR

This paper introduces Virtual Softmax, a simple method that injects a virtual negative class into the softmax function to enhance feature discriminability, leading to improved recognition performance.

Contribution

The paper proposes a novel Virtual Softmax technique that enlarges inter-class margins and compresses intra-class distributions by adding a virtual class, which is a new approach in discriminative feature learning.

Findings

Improves recognition accuracy on object classification tasks

Enhances face verification performance

Demonstrates superiority over existing methods

Abstract

Recently, learning discriminative features to improve the recognition performances gradually becomes the primary goal of deep learning, and numerous remarkable works have emerged. In this paper, we propose a novel yet extremely simple method \textbf{Virtual Softmax} to enhance the discriminative property of learned features by injecting a dynamic virtual negative class into the original softmax. Injecting virtual class aims to enlarge inter-class margin and compress intra-class distribution by strengthening the decision boundary constraint. Although it seems weird to optimize with this additional virtual class, we show that our method derives from an intuitive and clear motivation, and it indeed encourages the features to be more compact and separable. This paper empirically and experimentally demonstrates the superiority of Virtual Softmax, improving the performances on a variety of object classification and face verification tasks.