Call for Papers

Our workshop is nonarchival, the accepted papers will be posted on our website. Our workshop calls for high-quality and cutting-edge paper submissions in the following two tracks:

(A) Original Research Track

This track calls for 4-8 page paper (with unlimited references and appendices) of high-quality contributions from AI applications to all fields of scientific discovery, ranging from physics, biology, chemistry, earth science, environmental science, mechanical science, aerospace science, management science, agricultural science, material science, nuclear science etc. Appendix is optional, but reviewers are not required to read.

Example topics include (but not limited to):

• Learning from acoustics
• Learning physical dynamics from data
• Speeding up physical simulators, samplers and solvers
• Molecular modeling and de novo generation
• Modeling biological systems, genomics, protein, RNA
• Accelerating cosmological simulations
• Improving crop yields through precision agriculture
• Optimizing aerospace product design and development
• Benchmarking related or new tasks (i.e., datasets, sota models, etc.)
• Building tools/infrastructures/platforms for scientific discovery
• Study of science of science/scientific methods

(B) Attention Track

This track solicits 4-8 page paper (with unlimited references and appendices) that highlights a perspective of a subject in the field of AI for Science. We especially welcome contributions that discuss the gaps between AI and Science.

Example topics include (but not limited to):

• Unrealistic ML methodological assumptions
• Overlooked scientific questions
• Opportunities on the intersections of multiple disciplines
• Future research directions/hypothesis of an application area
• Responsible use and development of AI for science

(C) Highlight Track

This track solicits 4-8 page paper (with unlimited references and appendices) that is comprehensive survey/benchmark on a specific topic under AI4Science, e.g., ML for Molecules, comparing with the original track, this track is more focused on the are more interested in summarising the published works.

Example topics include (but not limited to):

• ML for molecule design
• ML for symbolic regression
• ML for combinatorial optimization
• ML for simulation

(D) Education Track

To support the ever-growing AI for Science field, education is an indispensable part of our community. We aim to solicit systematic and multi-level learning resources (including but not limited to courses, tutorials, notebooks, review papers, etc.) to bridge the educational gap in AI for Science. The content of the submission is flexible but a paper or report needs to be written using the LaTeX template. If the proposed content type is notebook with code example, we would still expect a short report to write about the overview and motivation about the topic and the learning resources for people who have further interest in the topic. As other tracks, we will highlight the best submissions from this track and invite the authors for a contributed talk at the workshop.

Specifically, the main goal/outcome of this track is to share the message and value of our AI for Science community:

Education is important; The community values education contribution; a platform for people passionate about education to connect. You may wonder AI for Science is so broad, whether your writing would be read by people and whether you should write about a specific field or a broader area or even the whole field. Our answer to this question is YES and we encourage ANY of them, that’s exactly why we open this track to collect expertise from the community and we may attempt to make some collective effort such as a roadmap of AI for Science to expand the impact and attract more people to contribute. We encourage researchers from diverse backgrounds to submit (in particular areas that did not attract enough attention).

Submission Instructions

Abstract submission is due on May 23th AoE, and paper submission is due on May 25th AoE. All submissions are managed through OpenReview.

The review process is double-blind so the submission should be anonymized. We welcome submissions that are (1) originally unpublished, (2) recently published, or (3) work-in-progress. Please use the ICML LaTeX template and change the title to under review at ICML 2024 AI for Science workshop. Accepted papers would be archived on the workshop website. Contributed talks and best paper awards would be given based on review scores and chairs discussion.