Two phases

You may be asked to review papers in either of our two phases (see Key Dates on the conference website). The review criteria in the two phases are the same. We would appreciate your suggestions for additional reviewers in both phases, either specific reviewers or general areas of expertise. (Of course, to suggest a specific reviewer, you should respect blind review, and put the suggestion in the “private comments to program committee” box.) The total number of papers you are asked to review across both phases will respect the limit we communicated to you during reviewer recruiting.

Review criteria

We are looking for creative, well-executed solutions to interesting problems throughout AI and statistics. Both incremental contributions and attempts at entirely new strategies for solving problems are welcome. But, the review process is known to favor incremental contributions over entirely new strategies, perhaps to an unfair degree; so, we ask reviewers to be aware of this effect, and to attempt to counter it in their own reviews.

In particular, we expect that there may be a tradeoff between “creativity” and “thoroughness”: creativity is the degree to which a paper represents a novel way of setting up a problem or an unusual approach to solving it, while thoroughness is the degree to which a paper supports its conclusions with extensive experimental or theoretical results. We are asking reviewers to rate both qualities separately, and to take into account this tradeoff when synthesizing their reviews into overall scores.

Simultaneous submissions

See the CFP for rules on simultaneous submissions. Briefly, other archival conferences are forbidden; tech reports and workshops without formal proceedings are OK, and journals are OK as long as their publication date is 2011 or later. If you suspect plagiarism or double submission, please let us know; the senior program committee will investigate, since they are allowed to know author names and can therefore evaluate the case better. Do not assume something is wrong: the authors may have notified the SPC of extenuating circumstances, and we have withheld such notifications from reviewers to allow double-blind reviewing.

Double-blind reviewing

See the submission page for the instructions we gave to authors. Please don't try to unblind a submission; but if you figure out the authors accidentally, or believe you may have figured out the authors, please let us know in the confidential program committee comments section of the review. If you believe that an accidental unblinding affects your ability to give a fair and unbiased review, please let us know as soon as possible, so that we can reassign the paper.

Supplementary material

For text supplementary material such as additional proofs, we've encouraged authors to include it at the end of the paper file. For text or other supplementary material, authors may also have provided a separate file. In either case you are not obliged to consult the supplementary material, but may do so if you believe it will help your review.

Expertise

All papers should be accessible to the general AISTATS audience. We would like your review even if there are parts of the paper where you feel you do not have the highest possible degree of expertise. Since we will often assign additional reviewers, we would appreciate your feedback about which areas of expertise would be helpful in reviewing the paper.