Directional statistics-based deep metric learning for image classification and retrieval. (September 2019)

• Record Type: Journal Article
• Authors: Zhe, Xuefei; Chen, Shifeng; Yan, Hong
• Is Part Of: Pattern recognition. Volume 93(2019:Sep.)
• Page Start: 113
• Page End: 123
• Publication Date: 2019-09
• Subjects: Deep distance metric learning -- Directional statistics -- Image retrieval -- Image similarity learning; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2019.04.005
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: Directional statistics is introduced to the deep distance metric learning. A simple but effective alternative learning algorithm is proposed to learning a robust discriminative hyper-spherical feature space. Significant improvements on benchmark datasets for classification and retrieval tasks. Higher classification accuracy can be reached with shallow convolutional neural networks. Abstract: L2-normalization is an effective method to enhance the discriminant power of deep representation learning. However, without exploiting the geometric properties of the feature space, the generally used gradient based optimization methods are failed to track the global information during training. In this paper, we propose a novel deep metric learning model based on the directional distribution. By defining the loss function based on the von Mises–Fisher distribution, we propose an effective alternative learning algorithm by periodically updating the class centers. The proposed metric learning not only captures the global information about the embedding space but also yields an approximate representation of the class distribution during training. Considering classification and retrieval tasks, our experiments on benchmark datasets demonstrate an improvement from the proposed algorithm. Particularly, with a small number of convolutional layers, a significant accuracy upsurge can be observed compared to widely used gradient-based methods.