NeurIPS 2020

Cycle-Contrast for Self-Supervised Video Representation Learning

Meta Review

The paper received mixed reviews. The reviewers noted that the idea is very interesting and novel, but at the same time, some comparisons are missing. The rebuttal clarified some of the concerns from the reviewers. While the paper would be stronger with additional baselines, the rebuttal clearly identifies why these baselines may be unnecessary. Papers do not need to obtain state-of-the-art results to be published, especially if they pursue interesting and alternative lines of research. The AC recommends the paper for acceptance.