Mixing Zero-Shot Learning Up: Learning Unseen Classes from Mixed Features

Zero-Shot Learning (ZSL) objective is to classify instances of classes that were not seen during the training phase. ZSL methods take advantage of side information, i.e., class attributes, to leverage information between the seen and unseen classes. Lately, generative methods have been used to synthesize unseen features in order to train a classifier for the unseen classes. Although generative methods obtain high performance, the learned distribution may not properly represent the real distribution of the unseen classes. We propose an approach to alleviate this issue by creating a new set of mixed features. These mixed features provided a new distribution for the generative method to learn from. By using these mixed features we obtained an +2.2% improvement over tf-VAEGAN in the Oxford Flowers (FLO) dataset.