Skip to yearly menu bar Skip to main content


Dear ICML participants --

 

COVID-19 is undoubtedly having an enormous impact on the world, and members of the ICML constituency--authors, reviewers, meta-reviewers, chairs, staff, and potential attendees--are no exception. After much deliberation, in conjunction with the ICML board and staff, and with input from ICLR 2020 and NeurIPS 2020, we have decided that ICML 2020 will take place entirely virtually. It is impossible to predict what the world will look like in July, and our first priority is the safety, health, and well-being of ICML participants.

Running a virtual conference introduces many challenges (especially when the change is made "in flight"), but we're enthusiastic that we will be able to provide a useful, meaningful, and intellectually stimulating conference experience, even though we will not be physically co-located. Moreover, along with the challenges, having a virtual conference also grants opportunities, and we are working to learn from, and expand upon, other conferences that are also shifting to virtual formats.

In addition, because so many of our lives are turned upside down right now, and many of us are scrambling to figure out our lives, schedules, and caring for family and loved ones, we have also decided to slow down the reviewing process to grant everyone more time to fit ICML responsibilities into their other responsibilities. Because the conference is virtual, we no longer need to hit a sufficiently early target for Visa reasons. Slowing the timeline was a very difficult decision to make, but we hope that doing so will strike a balance between keeping things moving and giving people an opportunity to re-plan their time with minimal harm to participants. We will shortly update the calendar on icml.cc to reflect these schedule changes.

 

What does this mean for you?

 

Authors: Papers will still be accepted and the proceedings will be produced as in all past years. Talks will be pre-recorded videos, with live Q&A sessions, more details to come. You will receive reviews somewhat later than originally planned (April 9), and will have more time to write your response (from April 9 through April 21). Accept/reject notifications are expected on June 1 (about a one month delay). We are working very hard to make sure you get at least three high-quality reviews, but ask for your patience in this difficult time.

We recognize that extending notifications introduces challenges for authors who would have wanted to submit a rejected ICML paper to another venue. In hopes of lessening this harm, we are adjusting (a) our dual-submission policy and (b) our withdrawal policy, as follows:

   (a) We will allow dual-submissions to other venues whose full paper submission deadline falls after 9 May 2020 (the original notification date). You will need to verify with the other venue that they will allow dual submissions.

   (b) We will allow authors to withdraw papers at any time between now and the end of the review process and papers withdrawn from today forward will not be considered in violation of dual-submission if you submit them elsewhere.

 

Reviewers: There is now a bit more time for reviewing. If you have not yet completed all of your reviewers but can complete them within one week (by March 30th at the latest), please email the respective meta-reviewer(s) to ask for an extension. Author response will come in later than originally planned (on April 21), and the discussion period will last a bit longer (from April 22 through May 5).

 

Meta-Reviewers: In addition to the aspects of the schedule that affect reviewers, we have also given you additional time to complete the meta-reviewing task. You now have until April 7 to ensure that all papers have at least three reviews, and to read over those to ensure they are polite and constructive (asking reviewers to make adjustments as needed). There is somewhat more time to make decision recommendations and write meta-reviews (due by May 15). You are welcome to begin the meta-review process early, any time after author-response has completed (on April 21).

 

(Virtual) Attendees: We are currently in the middle of planning the event, working closely with ICLR, following the lead of other conferences and also trying to invent some new mechanisms to take advantage of the virtual conference format. Registration fees will be substantially reduced and we hope that that, together with the lack of travel needs, will allow for many people to attend who may not have been able to otherwise. We will follow up with more information on icml.cc regarding concrete plans as we make them.

 

Sponsors: We are working on plans for how sponsors can participate (and recruit) as normal; updates will be available on the sponsor portal at icml.cc.

 

Everyone: One thing we'd ask for everyone in our community is kindness and patience. So many of us are under substantial unexpected stress, and we encourage everyone to support each other. The next steps of the paper review process, in particular, involve significant communication between all parties, and we ask everyone to understand that this is a difficult time for everyone and to assume best intentions on the parts of those who are working hard to make ICML a success.

 

 

We'd like to finish by expressing our sincere gratitude for everyone involved in the ICML process and in the process of figuring out how to make a virtual ICML 2020 a success. Anoush Najarian, Yoshua Bengio, and Kyunghyun Cho have all gone out of their way to help brainstorm ideas, and share information across conferences. Mary Ellen Perry and Lee Campbell have done--and will continue to do--an immense amount of work making all of this actually happen.

The ICML 2020 organizing committee--Alexandre Bouchard-Côté, John Cunningham, Marc Diesenroth, Barbara Engelhardt, Stefanie Jegelka, Varun Kanade, Emtiyas Khan, Samory Kpotufe, Simon Lacoste-Julien, Po-ling Loh, Matthias Seeger, Ivan Stelmakh, Alessandra Tosi, Caroline Uhler, Sinead Williamson, Andrew G. Wilson, Zhenyu Xue, and Jerry Zhu--has been instrumental in figuring out how to make this change mid-stream, and we thank them for this and all their upcoming work. And of course all of the program committee, especially the thousands of volunteer reviewers, are essential to making ICML a success and we truly value your commitment, time, and care.

 

Sincerely,

David Blei (ICML 2020 General Chair)
Hal Daumé III (ICML 2020 Program Co-Chair)
Aarti Singh (ICML 2020 Program Co-Chair)
John Langford (ICML Board President)