Diversity-boosted Generalization-Specialization Balancing for Zero-shot Learning

TL;DR

This paper introduces BGSNet, an end-to-end network that balances generalization and specialization in zero-shot learning by combining meta-learning, attentive feature extraction, and a novel diversity loss, leading to improved performance on benchmarks.

Contribution

The paper proposes BGSNet, a novel architecture that simultaneously balances generalization and specialization abilities in ZSL through innovative network design and loss functions.

Findings

BGSNet outperforms existing methods on four benchmark datasets.

The diversity loss effectively reduces redundancy and boosts feature diversity.

Component ablations confirm the importance of balancing generalization and specialization.

Abstract

Zero-Shot Learning (ZSL) aims to transfer classification capability from seen to unseen classes. Recent methods have proved that generalization and specialization are two essential abilities to achieve good performance in ZSL. However, focusing on only one of the abilities may result in models that are either too general with degraded classification ability or too specialized to generalize to unseen classes. In this paper, we propose an end-to-end network, termed as BGSNet, which equips and balances generalization and specialization abilities at the instance and dataset level. Specifically, BGSNet consists of two branches: the Generalization Network (GNet), which applies episodic meta-learning to learn generalized knowledge, and the Balanced Specialization Network (BSNet), which adopts multiple attentive extractors to extract discriminative features and achieve instance-level balance. A novel self-adjusted diversity loss is designed to optimize BSNet with redundancy reduced and diversity boosted. We further propose a differentiable dataset-level balance and update the weights in a linear annealing schedule to simulate network pruning and thus obtain the optimal structure for BSNet with dataset-level balance achieved. Experiments on four benchmark datasets demonstrate our model's effectiveness. Sufficient component ablations prove the necessity of integrating and balancing generalization and specialization abilities.