Application deadline: March 31 April 7, 2023, 23:59 Anywhere on Earth passed
Notification of acceptance: Early May, 2023
Note: Because of the ongoing war in Ukraine, we will not be able to provide visa invitation letters or travel grants to participants based in Russia.
The school welcomes applications from candidates all over the world, regardless of their level of expertise in Machine Learning or whether they are students or not; see FAQ.
To apply for attending the school, you need to provide the following information:
personal information (name, contact, affiliation etc.)
an up-to-date CV (your choice of template, must be a pdf file)
a statement of research interests (min 500 characters, max 2000 characters): describe your research interests, projects related to ML you worked on or would like to work on, motivation for attending the summer school.
an extended abstract on (choose ONLY ONE):
(1) [Research] your own project related to Machine Learning, OR
(2) [Reproduction] a project reproducing the results of a paper published by someone else in a top conference or journal (e.g. NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, AAAI, ACL, EMNLP, TPAMI, JMLR) in the past 5 years, OR
(3) [Review] a review of 3 related research papers on an ML topic of your choice (e.g. 3 papers on continual learning, or 3 papers on attention models, etc), published by other authors in top conferences or journals (e.g. NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, AAAI, ACL) OR
(4) [Competition] a report summarising your participation in an ML-related Kaggle competition or similar competition during 2021 / 2022 / 2023
The period for applications is closed.
Notes regarding the extended abstract
We provide four options for the extended abstract to make sure that the selection process is accessible to everyone, irrespective of the level of expertise in Machine Learning. We will assess interest and potential in the field, not just expertise in ML.
Accepted candidates who choose options (1) or (2) may be selected to present the projects as posters at the school, during the poster sessions.
The extended abstract should be structured as a short paper. E.g., it can have sections (e.g. Intro, Method, Experiments, References etc.), it can include images / figures / tables etc.
The extended abstract must not exceed two pages, excluding references.
Submissions are not blind: author(s) names and affiliations should appear on the first page.
It is possible to submit the extended abstract of a research paper that you've already published (from 2020 onward); this falls under option (1). Summarise your paper to respect the page limit, keep the original authors list, and include a brief description of your contribution to the paper if you are not the main author or shared main author; include this as a footnote attached to your name.
We accept multiple applicants (normally 2) with the same extended abstract for options (1) (2) or (4), provided that the applicants all contributed equally or considerably to the work. In your extended abstract, put the names of all the applicants and include a brief description of your contribution to the project, as a footnote attached to your name. Note that the selection decision takes into account several factors, not only the abstract, hence it can differ for candidates with the same abstract.
All documents must be submitted in pdf format.
Best poster awards will be selected during the school.
Recommendations for choosing among the four options of extended abstract:
If you are actively working on ML research projects (as a PhD, postdoc, Master student, faculty, practitioner in ML etc.), you should choose option (1) or (2).
If you are a beginner in ML, options (3) or (4) could be more accessible.
For option (1), it is fine if the work is in incipient stages and you only have preliminary results. This is a non-archival submission; you can submit your work to any other venue as well.
For option (2), please also include, if possible, your own observations about the paper, e.g. analysis of robustness, was the paper difficult to reproduce given the details in the paper; most important tricks in implementation etc.
For option (3), consider reviewing works that have some aspects in common, so that you can compare them and highlight their pros and cons.
For option (4), describe the problem, the data, and the algorithms you implemented for your challenge submission.
In the application form, you will need to indicate the option you chose. Don't hesitate to contact us at contact at eeml dot eu if you are unsure about which type of abstract to select.
Evaluation of the abstracts will be based on the aspects below:
Clarity of presentation (good writing, includes figures, etc)
Relevance for the school theme
Properly contextualising the work within the literature (discussing briefly related work)
Experiments if any, interpretation of the results, and conclusion
Clarity of the description of the paper reproduced
Understanding of the paper reproduced
Clear description of the reproduction setting / implementation
Discuss results of the reproduction
Comments on the difficulty of reproducing results / criticisms
(bonus) Further exploration of the method, e.g. robustness to hyper-parameters, additional testing on a different dataset, etc.
Clear description of the reviewed papers
Understanding of the reviewed papers
Comparison between the 3 papers, highlight advantages and downsides for each
(bonus) Provide additional interpretation / insights on the works, or re-interpret one work in the light of the other
Clear description of the problem and data (provided by the challenge organisers or extra data you used)
Detailed description of the solution
Mention of the ranking obtained
Discussion of challenges encountered during the implementation
Reasons for rejection:
The extended abstract is a blank pdf
The extended abstract is a one-paragraph text (half a page or less)
The extended abstract is the same as the CV
The extended abstract goes beyond the page limit
Please contact us at contact at eeml dot eu if you have any questions regarding your choice of abstract, or the application process in general.