From Classical to Generalized Zero-Shot Learning: a Simple Adaptation Process

TL;DR

This paper introduces a simple adaptation process for generalized zero-shot learning that improves performance on both seen and unseen classes by better hyper-parameter selection, applicable to existing methods.

Contribution

It proposes a new training and evaluation process for GZSL that enhances existing approaches without altering their core algorithms.

Findings

GZSL performance improved from 28.5 to 42.2 on CUB dataset.

GZSL performance improved from 28.2 to 57.1 on AwA2 dataset.

The process is applicable to any existing ZSL approach.

Abstract

Zero-shot learning (ZSL) is concerned with the recognition of previously unseen classes. It relies on additional semantic knowledge for which a mapping can be learned with training examples of seen classes. While classical ZSL considers the recognition performance on unseen classes only, generalized zero-shot learning (GZSL) aims at maximizing performance on both seen and unseen classes. In this paper, we propose a new process for training and evaluation in the GZSL setting; this process addresses the gap in performance between samples from unseen and seen classes by penalizing the latter, and enables to select hyper-parameters well-suited to the GZSL task. It can be applied to any existing ZSL approach and leads to a significant performance boost: the experimental evaluation shows that GZSL performance, averaged over eight state-of-the-art methods, is improved from 28.5 to 42.2 on CUB and from 28.2 to 57.1 on AwA2.