We (the program committee: Zico Kolter, Katherine Heller, Adrian Weller, and Nuria Oliver) are thrilled to have recently put out the ICML 2024 call for papers. While this official call documents the details of the submission process, there are a number of new elements and aspects to this year’s ICML process that we want to specifically call your attention to. To do this, we are planning a series of blog posts that we will post under this account, highlighting some of our decisions for this year and the philosophy behind them. We welcome and encourage any feedback on these topics. Machine learning conferences are constantly trying out new ideas, evolving, and we rely on your feedback and comments in order to improve going forward.

First off, before getting into what is new this year, we should mention that most of the process this year at ICML closely follows what was done in previous years. The deadline for paper submissions is February 1st, Anywhere on Earth (11:59:59pm UTC-12). Adopting last year’s changes, we will be using the OpenReview system for reviews (papers will still be private, available only to the reviewers during the review period, maintaining the practice at previous ICML and NeurIPS conferences).

But there are a number of new aspects this year, which we briefly highlight below, andwhich we will also devote separate posts to in the coming weeks.

Position paper track
A notable new aspect of the call this year is that in addition to technical papers, we are soliciting submissions to a separate position paper track at the conference. As machine learning has grown as a field, many of the most important discussions are happening outside of traditional technical papers. To encourage discussion of these topics, and respectful exchanges of views, we are forming a separate track specifically for papers that advocate for a particular viewpoint or try to increase the discussion and exposure of a certain perspective. Papers submitted to this track will be reviewed within the normal process according to clear criteria.

Reviewer and AC recruiting
The ICML reviewing process will be organized, as usual, into reviewers, area chairs (ACs), and senior area chairs (SACs). There will be the normal initial reviewing process, followed by a one-week rebuttal period, and a one-week discussion period. This year, in addition to the typical reviewer/SAC/AC recruiting process, everyone will have the option to nominate themselves or others for these roles. The nomination form will open on November 13, accompanied by another blog post providing more detail on the process.

Ethics Review
Bringing ICML more in line with the practice at ICLR and NeurIPS, this year we are initiating an ethics review committee for ICML papers. We will detail this process in a later post, but one item to note is that including an ethics statement is required on paper submissions (this does not count toward the page limit). We provide a template response for papers where the authors believe there to be no explicit ethics implications beyond the general scope of those applied to machine learning in general.

Comments and Feedback
We’d love to hear your thoughts. If you have any comments, suggestions or feedback, please email us at [email protected], or leave a comment below.