TGG: Transferable Graph Generation for Zero-shot and Few-shot Learning

TL;DR

This paper introduces TGG, a graph-based framework that explicitly models and transfers relations between seen and unseen classes to improve zero-shot and few-shot learning, addressing domain shift and data scarcity issues.

Contribution

The paper proposes a novel Transferable Graph Generation approach that explicitly models relations via graph generation and dual relation propagation, unifying zero-shot, generalized zero-shot, and few-shot learning.

Findings

TGG outperforms existing methods significantly across multiple benchmarks.

The explicit relation modeling improves knowledge transfer to unseen classes.

Dual relation propagation effectively mitigates domain shift.

Abstract

Zero-shot and few-shot learning aim to improve generalization to unseen concepts, which are promising in many realistic scenarios. Due to the lack of data in unseen domain, relation modeling between seen and unseen domains is vital for knowledge transfer in these tasks. Most existing methods capture seen-unseen relation implicitly via semantic embedding or feature generation, resulting in inadequate use of relation and some issues remain (e.g. domain shift). To tackle these challenges, we propose a Transferable Graph Generation (TGG) approach, in which the relation is modeled and utilized explicitly via graph generation. Specifically, our proposed TGG contains two main components: (1) Graph generation for relation modeling. An attention-based aggregate network and a relation kernel are proposed, which generate instance-level graph based on a class-level prototype graph and visual features. Proximity information aggregating is guided by a multi-head graph attention mechanism, where seen and unseen features synthesized by GAN are revised as node embeddings. The relation kernel further generates edges with GCN and graph kernel method, to capture instance-level topological structure while tackling data imbalance and noise. (2) Relation propagation for relation utilization. A dual relation propagation approach is proposed, where relations captured by the generated graph are separately propagated from the seen and unseen subgraphs. The two propagations learn from each other in a dual learning fashion, which performs as an adaptation way for mitigating domain shift. All components are jointly optimized with a meta-learning strategy, and our TGG acts as an end-to-end framework unifying conventional zero-shot, generalized zero-shot and few-shot learning. Extensive experiments demonstrate that it consistently surpasses existing methods of the above three fields by a significant margin.