Deconstructed Generation-Based Zero-Shot Model

TL;DR

This paper deconstructs the generator-classifier framework in GZSL, highlighting the importance of attribute generalization and independent classifier learning, and proposes a simple, effective method that outperforms existing models without relying on generative models.

Contribution

It provides a fundamental analysis of generator-based GZSL methods, introduces a simplified approach emphasizing attribute generalization and classifier independence, and achieves superior results on multiple datasets.

Findings

Outperforms SotAs on four GZSL datasets

Effective even without generative models

Highlights importance of attribute generalization

Abstract

Recent research on Generalized Zero-Shot Learning (GZSL) has focused primarily on generation-based methods. However, current literature has overlooked the fundamental principles of these methods and has made limited progress in a complex manner. In this paper, we aim to deconstruct the generator-classifier framework and provide guidance for its improvement and extension. We begin by breaking down the generator-learned unseen class distribution into class-level and instance-level distributions. Through our analysis of the role of these two types of distributions in solving the GZSL problem, we generalize the focus of the generation-based approach, emphasizing the importance of (i) attribute generalization in generator learning and (ii) independent classifier learning with partially biased data. We present a simple method based on this analysis that outperforms SotAs on four public GZSL datasets, demonstrating the validity of our deconstruction. Furthermore, our proposed method remains effective even without a generative model, representing a step towards simplifying the generator-classifier structure. Our code is available at \url{https://github.com/cdb342/DGZ}.