NeurIPS 2020

What went wrong and when? Instance-wise feature importance for time-series black-box models

Meta Review

Three knowledgeable referees support acceptance for the contributions, notably R1, R3 and R4. After the discussion, R2 does not oppose acceptance and I don't see any major criticism to the paper. Therefore, I also recommend acceptance. However, please consider revising your paper to address a common comment by the reviewers regarding non-instantaneous attribution.