NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:7817
Title:Defending Neural Backdoors via Generative Distribution Modeling

Thanks for the paper submission and for addressing concerns in the reviews. We believe this is a valuable contribution to NeurIPS. The main contribution on backdoor defenses for neural networks is interesting and may open the door to further research in this direction. As areas of improvement, we strongly recommend simplifying the densest parts of the presentation (as long as completeness is maintained), enhance the related work (as addressed in the author feedback), and class equivalency (similarly trigger choice). Please refer to the detailed feedback for valuable suggestions and comments from the reviewers.