/aistats-review-process

name: aistats-review-process
description: Use when explaining or planning around AISTATS peer review, OpenReview review release, author-reviewer discussion, reviewer volunteer expectations, reviewer confidentiality, decision criteria, meta-review dynamics, the statistician-heavy reviewer pool, and PMLR proceedings outcomes.

AISTATS Review Process

Use this to reason about review-stage strategy. Reopen the current CFP, OpenReview group, author instructions, reviewer instructions if posted, and code of conduct before making process claims.

Process model

• AISTATS uses OpenReview for submission and review workflow in recent cycles.
• Reviewers evaluate technical correctness, statistical and machine-learning contribution,

empirical support, clarity, reproducibility, and relevance to artificial intelligence and statistics.

• Author discussion is limited. AISTATS 2026 used a discussion period after initial reviews,

with text-only author-reviewer discussion and no links.

• Reviewer and author obligations include confidentiality, appropriate conflicts,

professional conduct, and respect for anonymity.

• The most useful response is a decision-focused clarification that gives the area chair or

meta-reviewer a clean rationale for acceptance or rejection.

• Accepted papers are published in PMLR, so final metadata and camera-ready compliance matter

as much as the initial acceptance.

Who reviews here

• The pool mixes ML researchers with statisticians and statistical learning theorists;

expect at least one reviewer to read proofs and assumption sets line by line.

• Because AISTATS is smaller and more specialized than NeurIPS or ICML, topical matches are

closer, so vague proof sketches get caught rather than skimmed past.

• Borderline theory-plus-experiments papers usually fall on one of three edges: an assumption

the experiments do not satisfy, a missing classical-statistics baseline, or a rate claim never checked empirically.

Scoring leverage table

| Review dimension | What raises it | What sinks it | |---|---|---| | Correctness | Complete assumption statements with a main-text proof sketch | Hidden conditions; constants swept into O-notation when they matter | | Significance | A guarantee the ML literature lacked, or a practical method statistics lacked | Incremental rate gain with no conceptual or practical payoff | | Empirical support | Experiments engineered to probe the theory | Benchmarks disconnected from the theorem regimes | | Clarity | Numbered assumptions and a single notation source | Notation collisions between sections |

Stage-by-stage realism

• Initial reviews: triage by what the meta-reviewer would weigh, not by reviewer tone.
• Discussion: windows are short; an early, precise reply is worth more than a late

comprehensive one.

• Decision: the meta-review synthesizes; one unanswered correctness objection outweighs

several resolved clarity complaints.

• Reviewer-volunteer expectations for submitting authors have appeared in recent cycles;

confirm the current CFP rather than assuming either way.

Output format

[Current stage] submitted / reviews / discussion / decision / camera-ready
[Decision actors] <reviewers/meta-reviewer/chairs>
[Likely leverage] <correctness/statistics/experiments/clarity/reproducibility>
[Forbidden moves] <identity leak / external links if forbidden / new unsupported results>
[Next response move] <one action>