<div dir="ltr"><div>Thanks, that was a nice post, I enjoyed it. It struck me as related to the FARR project, which is trying to tackle a similar set of issues as those you outline:</div><div><br></div><div>FARR: FAIR in ML, AI Readiness, & Reproducibility (<a href="https://www.gofair.us/farr">https://www.gofair.us/farr</a>)</div><div><br></div><div>They published a recent case study on publishing AI models in physics, and are working on other disciplinary issues as well:</div><div><br></div><div><div class="gmail-csl-bib-body" style="line-height:1.35">
  <div class="gmail-csl-entry">Duarte J, Li H, Roy A, Zhu R, Huerta EA, Diaz D, Harris P, Kansal R, Katz DS, Kavoori IH, Kindratenko VV, Mokhtar F, Neubauer MS, Park SE, Quinnan M, Rusack R, Zhao Z (2022) FAIR AI Models in High Energy Physics. <a href="https://doi.org/10.48550/arXiv.2212.05081">https://doi.org/10.48550/arXiv.2212.05081</a></div>
  <span class="gmail-Z3988" title="url_ver=Z39.88-2004&ctx_ver=Z39.88-2004&rfr_id=info%3Asid%2Fzotero.org%3A2&rft_id=info%3Adoi%2F10.48550%2FarXiv.2212.05081&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.type=preprint&rft.title=FAIR%20AI%20Models%20in%20High%20Energy%20Physics&rft.description=The%20findable%2C%20accessible%2C%20interoperable%2C%20and%20reusable%20(FAIR)%20data%20principles%20have%20provided%20a%20framework%20for%20examining%2C%20evaluating%2C%20and%20improving%20how%20we%20share%20data%20with%20the%20aim%20of%20facilitating%20scientific%20discovery.%20Efforts%20have%20been%20made%20to%20generalize%20these%20principles%20to%20research%20software%20and%20other%20digital%20products.%20Artificial%20intelligence%20(AI)%20models%20--%20algorithms%20that%20have%20been%20trained%20on%20data%20rather%20than%20explicitly%20programmed%20--%20are%20an%20important%20target%20for%20this%20because%20of%20the%20ever-increasing%20pace%20with%20which%20AI%20is%20transforming%20scientific%20and%20engineering%20domains.%20In%20this%20paper%2C%20we%20propose%20a%20practical%20definition%20of%20FAIR%20principles%20for%20AI%20models%20and%20create%20a%20FAIR%20AI%20project%20template%20that%20promotes%20adherence%20to%20these%20principles.%20We%20demonstrate%20how%20to%20implement%20these%20principles%20using%20a%20concrete%20example%20from%20experimental%20high%20energy%20physics%3A%20a%20graph%20neural%20network%20for%20identifying%20Higgs%20bosons%20decaying%20to%20bottom%20quarks.%20We%20study%20the%20robustness%20of%20these%20FAIR%20AI%20models%20and%20their%20portability%20across%20hardware%20architectures%20and%20software%20frameworks%2C%20and%20report%20new%20insights%20on%20the%20interpretability%20of%20AI%20predictions%20by%20studying%20the%20interplay%20between%20FAIR%20datasets%20and%20AI%20models.%20Enabled%20by%20publishing%20FAIR%20AI%20models%2C%20these%20studies%20pave%20the%20way%20toward%20reliable%20and%20automated%20AI-driven%20scientific%20discovery.&rft.identifier=urn%3Adoi%3A10.48550%2FarXiv.2212.05081&rft.aufirst=Javier&rft.aulast=Duarte&rft.au=Javier%20Duarte&rft.au=Haoyang%20Li&rft.au=Avik%20Roy&rft.au=Ruike%20Zhu&rft.au=E.%20A.%20Huerta&rft.au=Daniel%20Diaz&rft.au=Philip%20Harris&rft.au=Raghav%20Kansal&rft.au=Daniel%20S.%20Katz&rft.au=Ishaan%20H.%20Kavoori&rft.au=Volodymyr%20V.%20Kindratenko&rft.au=Farouk%20Mokhtar&rft.au=Mark%20S.%20Neubauer&rft.au=Sang%20Eon%20Park&rft.au=Melissa%20Quinnan&rft.au=Roger%20Rusack&rft.au=Zhizhen%20Zhao&rft.date=2022-12-21"></span> <br></div><div class="gmail-csl-bib-body" style="line-height:1.35">Matt<br></div></div><div><br></div><div><div><div dir="ltr" class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div><b>Matthew B. Jones</b></div><div>ORCID: <a href="https://orcid.org/0000-0003-0077-4738" target="_blank">0000-0003-0077-4738</a></div><div>
Director of Informatics R&D, <a href="http://www.nceas.ucsb.edu/ecoinfo" style="color:rgb(17,85,204)" target="_blank">National Center for Ecological Analysis and Synthesis</a></div><div>PI, NSF <a href="https://arcticdata.io/" style="color:rgb(17,85,204)" target="_blank">Arctic Data Center</a></div><div>Director, <a href="https://dataone.org/" style="color:rgb(17,85,204)" target="_blank">DataONE</a> program
</div><div>
University of California Santa Barbara</div></div></div></div></div></div><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Apr 25, 2023 at 7:01 AM Pascal Heus <<a href="mailto:pascal.heus@postman.com">pascal.heus@postman.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">Dear colleagues:<div>I just published this short article around AI transparency and metadata challenges which might be of interest to you. One of the objectives is to foster a bridge between the AI and (meta)data communities.</div><div><a href="https://plgah.medium.com/ai-has-a-metadata-problem-78b30ca1936b" target="_blank">https://plgah.medium.com/ai-has-a-metadata-problem-78b30ca1936b</a><br></div><div>Thoughts, feedback, suggestions most appreciated.</div><div>Best,</div><div>*P</div></div>
_______________________________________________<br>
CODATA-international mailing list<br>
<a href="mailto:CODATA-international@lists.codata.org" target="_blank">CODATA-international@lists.codata.org</a><br>
<a href="http://lists.codata.org/mailman/listinfo/codata-international_lists.codata.org" rel="noreferrer" target="_blank">http://lists.codata.org/mailman/listinfo/codata-international_lists.codata.org</a><br>
<br>
The CODATA International list is for announcement of of activities, events and outputs by CODATA and by other organisations and initiatives.  It is also for discussion of all issues related to data.  It is an open subscription list with only lightweight moderation to remove spam.  Messages posted on the list by third parties do not necessarily imply endorsement by CODATA.<br>
</blockquote></div>