NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:36
Title:Zero-shot Learning via Simultaneous Generating and Learning

The addresses zero-shot learning by an EM process of iteratively generating examples from unseen classes, and learning with them, this leads to generating samples that are good to learn from. Reviewers found the idea novel for this context, the writing clear and the experiments (mostly) convincing. They asked to see additional ablation studies in the final version.