Leveraging Seen and Unseen Semantic Relationships for Generative Zero-Shot Learning

TL;DR

This paper introduces LsrGAN, a generative model that explicitly leverages semantic relationships between seen and unseen classes to improve zero-shot learning performance, especially in generalized settings.

Contribution

LsrGAN incorporates a novel Semantic Regularized Loss to explicitly transfer knowledge between seen and unseen classes, addressing overfitting issues in generative ZSL models.

Findings

LsrGAN outperforms previous methods on seven benchmark datasets.

The model performs well on both ZSL and GZSL tasks.

Semantic regularization improves the quality of generated features.

Abstract

Zero-shot learning (ZSL) addresses the unseen class recognition problem by leveraging semantic information to transfer knowledge from seen classes to unseen classes. Generative models synthesize the unseen visual features and convert ZSL into a classical supervised learning problem. These generative models are trained using the seen classes and are expected to implicitly transfer the knowledge from seen to unseen classes. However, their performance is stymied by overfitting, which leads to substandard performance on Generalized Zero-Shot learning (GZSL). To address this concern, we propose the novel LsrGAN, a generative model that Leverages the Semantic Relationship between seen and unseen categories and explicitly performs knowledge transfer by incorporating a novel Semantic Regularized Loss (SR-Loss). The SR-loss guides the LsrGAN to generate visual features that mirror the semantic relationships between seen and unseen classes. Experiments on seven benchmark datasets, including the challenging Wikipedia text-based CUB and NABirds splits, and Attribute-based AWA, CUB, and SUN, demonstrates the superiority of the LsrGAN compared to previous state-of-the-art approaches under both ZSL and GZSL. Code is available at https: // github. com/ Maunil/ LsrGAN