Evolutionary Generalized Zero-Shot Learning

TL;DR

This paper introduces Evolutionary Generalized Zero-Shot Learning (EGZSL), a method enabling models to adapt and evolve online from test data streams, addressing challenges like forgetting and bias, demonstrated on benchmark datasets.

Contribution

It proposes a novel online evolutionary framework for zero-shot learning that continuously adapts from unlabeled test data, overcoming key challenges and outperforming baselines.

Findings

Model learns from test data stream successfully.

Outperforms baseline methods on benchmark datasets.

Addresses catastrophic forgetting and bias in online zero-shot learning.

Abstract

Attribute-based Zero-Shot Learning (ZSL) has revolutionized the ability of models to recognize new classes not seen during training. However, with the advancement of large-scale models, the expectations have risen. Beyond merely achieving zero-shot generalization, there is a growing demand for universal models that can continually evolve in expert domains using unlabeled data. To address this, we introduce a scaled-down instantiation of this challenge: Evolutionary Generalized Zero-Shot Learning (EGZSL). This setting allows a low-performing zero-shot model to adapt to the test data stream and evolve online. We elaborate on three challenges of this special task, \ie, catastrophic forgetting, initial prediction bias, and evolutionary data class bias. Moreover, we propose targeted solutions for each challenge, resulting in a generic method capable of continuous evolution from a given initial IGZSL model. Experiments on three popular GZSL benchmark datasets demonstrate that our model can learn from the test data stream while other baselines fail. Codes are available at \url{https://github.com/cdb342/EGZSL}.