Bi-directional Distribution Alignment for Transductive Zero-Shot Learning

Zero-shot learning (ZSL) suffers intensely from the domain shift issue, i.e., the mismatch (or misalignment) between the true and learned data distributions for classes without training data (unseen classes). By learning additionally from unlabelled data collected for the unseen classes, transductive ZSL (TZSL) could reduce the shift but only to a certain extent. To improve TZSL, we propose a novel approach Bi-VAEGAN which strengthens the distribution alignment between the visual space and an auxiliary space. As a result, it can reduce largely the domain shift. The proposed key designs include (1) a bi-directional distribution alignment, (2) a simple but effective L-2-norm based feature normalization approach, and (3) a more sophisticated unseen class prior estimation. Evaluated by four benchmark datasets, Bi-VAEGAN(1) achieves the new state of the art under both the standard and generalized TZSL settings.