Unseen Classes at a Later Time? No Problem

TL;DR

This paper introduces a new practical setting for continual generalized zero-shot learning (CGZSL) and proposes a unified feature-generative framework that dynamically adapts to new classes over time, outperforming existing methods.

Contribution

The paper consolidates CGZSL variants, proposes the Online-CGZSL setting, and introduces a bi-directional incremental alignment framework for dynamic class addition.

Findings

Our approach outperforms baselines on five benchmark datasets.

It performs especially well in the practical Online-CGZSL setting.

The method effectively handles addition of new classes with or without labeled data.

Abstract

Recent progress towards learning from limited supervision has encouraged efforts towards designing models that can recognize novel classes at test time (generalized zero-shot learning or GZSL). GZSL approaches assume knowledge of all classes, with or without labeled data, beforehand. However, practical scenarios demand models that are adaptable and can handle dynamic addition of new seen and unseen classes on the fly (that is continual generalized zero-shot learning or CGZSL). One solution is to sequentially retrain and reuse conventional GZSL methods, however, such an approach suffers from catastrophic forgetting leading to suboptimal generalization performance. A few recent efforts towards tackling CGZSL have been limited by difference in settings, practicality, data splits and protocols followed-inhibiting fair comparison and a clear direction forward. Motivated from these observations, in this work, we firstly consolidate the different CGZSL setting variants and propose a new Online-CGZSL setting which is more practical and flexible. Secondly, we introduce a unified feature-generative framework for CGZSL that leverages bi-directional incremental alignment to dynamically adapt to addition of new classes, with or without labeled data, that arrive over time in any of these CGZSL settings. Our comprehensive experiments and analysis on five benchmark datasets and comparison with baselines show that our approach consistently outperforms existing methods, especially on the more practical Online setting.