Transductive Zero-Shot Learning by Decoupled Feature Generation

TL;DR

This paper introduces a decoupled generative approach for transductive zero-shot learning, improving the synthesis of visual features by separately modeling data distribution and semantic translation, leading to superior performance.

Contribution

It proposes a novel decoupling strategy for feature generation in ZSL, separating data distribution modeling from semantic translation, which enhances synthesis quality and accuracy.

Findings

Decoupled model outperforms state-of-the-art methods.

Separate generators improve feature realism and semantic alignment.

Ablation study confirms the effectiveness of decoupling.

Abstract

In this paper, we address zero-shot learning (ZSL), the problem of recognizing categories for which no labeled visual data are available during training. We focus on the transductive setting, in which unlabelled visual data from unseen classes is available. State-of-the-art paradigms in ZSL typically exploit generative adversarial networks to synthesize visual features from semantic attributes. We posit that the main limitation of these approaches is to adopt a single model to face two problems: 1) generating realistic visual features, and 2) translating semantic attributes into visual cues. Differently, we propose to decouple such tasks, solving them separately. In particular, we train an unconditional generator to solely capture the complexity of the distribution of visual data and we subsequently pair it with a conditional generator devoted to enrich the prior knowledge of the data distribution with the semantic content of the class embeddings. We present a detailed ablation study to dissect the effect of our proposed decoupling approach, while demonstrating its superiority over the related state-of-the-art.