Journal
KNOWLEDGE-BASED SYSTEMS
Volume 215, Issue -, Pages -Publisher
ELSEVIER
DOI: 10.1016/j.knosys.2021.106773
Keywords
Zero-shot learning; Generalized zero-shot learning; Orthogonal projection
Categories
Funding
- Hong Kong Research Grant Council (HKRGC) [GRF 12200317, 12300218, 12300519, 17201020]
- HKU-TCL Joint Research Centre for Artificial Intelligence, Hong Kong
- National Natural Science Foundation of China (NSFC) [61876208]
- Key-Area Research and Development Program of Guangdong Province, China [2018B010108002]
- Pearl River S&T Nova Program of Guangzhou, China [201806010081]
Ask authors/readers for more resources
This paper proposes a method to match latent visual and semantic representations by exploiting shared concepts in a common subspace, while introducing reconstruction losses for both types of features to reduce domain shift and information loss. Experimental results demonstrate that the accuracy of the proposed method outperforms existing ZSL methods on six benchmark datasets.
Zero-shot learning (ZSL) aims to classify instances whose classes could be unseen during training. Most existing ZSL methods project visual or semantic features into the space of the other one, or into a common subspace. The main goal of projection is to find out the similar features in the latent subspace. However, existing methods barely consider common features that preserve knowledge, here we refer to these features as the shared concepts, which are essential to model the relationship between the visual and semantic spaces. In this paper, we exploit the underlying concepts shared by both visual and semantic features in a latent common subspace and propose to match their latent visual and semantic representations. To reduce domain shift and information loss, we introduce reconstruction losses for both visual and semantic features. As a result, the reconstruction regularizations are added to the similar features and thereby obtain knowledge preserving shared concepts via the proposed method. Mathematically, it is formulated as the minimization problem for mutual orthogonal projection to their latent common subspace. The problem involves two projection variables, thus we develop an algorithm based on the Gauss-Seidel iteration scheme and split the problem into two subproblems in the scheme. These two subproblems are further solved by searching algorithms based on the Barzilai-Borwein stepsize. Extensive experiments on six benchmark data sets are conducted to demonstrate that the accuracy of the proposed method is better than that of existing ZSL methods. (C) 2021 Elsevier B.V. All rights reserved.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available