Transductive Zero-Shot Learning by Decoupled Feature Generation
This work aims to improve zero-shot learning performance for researchers working on recognizing categories without labeled visual data, representing an incremental improvement in methodology.
This paper addresses transductive zero-shot learning by decoupling the generation of realistic visual features from the translation of semantic attributes into visual cues. They propose training an unconditional generator for visual data distribution and a conditional generator for semantic content, demonstrating superiority over state-of-the-art methods.
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.