NeurIPS 2020

Robust Multi-Object Matching via Iterative Reweighting of the Graph Connection Laplacian

Meta Review

This paper generated quite a bit of discussion. While three of the reviewers found the paper to be above the acceptance threshold, the fourth one was concerned about the clarity of the writing and the relationship to IRLS. In the rebuttal, the authors called R4's review a "an outlier" but did a good job of responding to comments from the other three reviewers. I am happy to recommend accepting this paper but ask that the authors prepare a significantly revised version that addresses the issues raised by the reviewers. I should point out that the concerns raised by R4 were supported by the other reviewers as well during discussion, hence he/she is not an outlier.