Robust Bidirectional Generative Network for Generalized Zero-shot Learning

In this work, we propose a novel generative approach named Robust Bidirectional Generative Network (RBGN) based on Conditional Generative Adversarial Network (CGAN) for Generalized Zero-shot Learning (GZSL). RBGN employs the adversarial attack to train a more rigorous discriminator, and then enhances the generalization ability and robustness of the feature generator under minimax strategy. Moreover, RBGN decodes the generated visual features back to their semantic representations through visual to semantic decoding network for further improving the representational ability of generated visual features and alleviating the hubness problem. The experimental results on four datasets, i.e. CUB, SUN, AWA1, AWA2, demonstrate that our model achieves competitive performance in comparison with the state-of-the-art approaches and owns better generalization ability to unseen classes over the conventional generative GZSL models. The further robustness analysis also validates the strong robustness of our model to the different types of semantic disturbance.