Papers go to venues such as NeurIPS, ICML, ICLR and ACL.
Peer review is double-blind at most major machine learning conferences.
arXiv preprints share work immediately, before or during review.
Releasing code and data is now effectively expected.
Rejection is routine, and resubmitting after revision is the normal path.
Machine learning research is published mainly at conferences rather than in journals.
NeurIPS, ICML and ICLR are the general venues, with ACL for language and CVPR for vision.
Each has a fixed deadline, which shapes the rhythm of the field.
Review is double-blind at most major venues, so submissions are anonymised.
Three or four reviewers score the paper and a discussion period follows.
Authors get a rebuttal, which occasionally changes a score.
Acceptance rates sit around twenty to thirty per cent, so rejection is the normal outcome.
Reviews vary widely in quality, and a bad review is not evidence of a bad paper.
Revising and resubmitting to the next venue is the standard path.
arXiv preprints share work immediately, before or during review.
That is now normal and is how most people actually read new work.
Releasing code and data is effectively expected, and its absence is noticed.
A reproducibility checklist is part of submission at several venues.
Workshops are a lower-stakes way to get early work in front of people.
Citations accumulate slowly, and the paper that matters is rarely the one you expected.
One question on this card. Every option explains itself once you pick, right or wrong.
11 concepts in this module, then a graded test with every answer explained.