In 1 Hour! @afedercooper @yuchengthekid and I will be at posters with "Hyperparameter Optimization Is Deceiving Us, and How to Stop It" Come chat about how we draw conclusions from experiments in ML & how we can make our processes more rigorous hyperparameter-deception.github.io

Happening now! We're in G3 in the optimization poster room by the purple demon 😈 @afedercooper NeurIPS Conference neurips.cc/virtual/2021/p…

Thank you everyone (attendees, presenters, speakers, reviewers, volunteers and organizers) for making ICBINB@NeurIPS 2021 a success!! We would love to hear back from you all, please fill this 3min survey: forms.gle/boR2JysufkFkPv…

Happy to share that we'll be presenting a tutorial about reproducibility in NLP at #ACL2022! This will be a joint effort between UvA Amsterdam and The ML Reproducibility Challenge, CC: @MauritsBleeker, Samarth Bhargav, Jessica Zosa Forde, Koustuv Sinha ✈️ NeurIPS, Jesse Dodge, Sasha Luccioni, PhD 🦋🌎✨🤗 and Robert Stojnic! 🥳🥳🥳

It is official! We have launched the ICBINB initiative! (icbinb.cc) Let's crack open the research process together. Thank you Aaron Schein @franciscuto _hylandSL - not here Jessica Zosa Forde Max Welling Tamara Broderick and Bob Williamson to make it happen!!

Excited to host Tamara Broderick (MIT) for the 1st ICBINB monthly seminar series!! When: Thursday, Feb 3rd at 10:00AM EST. Where: RSVP for the Zoom link: icbinb.cc/seminar-series/ Title: An Automatic Finite-Sample Robustness Metric: Can Dropping a Little Data Change Conclusions?

Our #FAccT2022 #CRAFT session on *Communication Across Communities in Machine Learning Research and Practice*! w @stefesseM Adriana Romero-Soriano Caglar Gulcehre Levent Sagun seyi Samuel Bell Jessica Zosa Forde sites.google.com/view/sedl-work…

I’ll be co-presenting a tutorial (remotely) on Reproducibility in NLP at #acl2022 tomorrow 9.30AM Dublin time, with Ana Lučić Jesse Dodge @MauritsBleeker Samarth Bhargav Jessica Forde & Robert Stojnic. Details : 2022.aclweb.org/tutorials. See you there!

Join us at the ACL tutorial on reproducibility happening now! underline.io/events/284/ses… with Ana Lučić Jessica Zosa Forde Jesse Dodge Koustuv Sinha ✈️ NeurIPS Sasha Luccioni, PhD 🦋🌎✨🤗 @MauritsBleeker Samarth Bhargav Robert Stojnic

Come check out @MauritsBleeker presenting about the FACT-AI course from UvA Amsterdam at our #ACL2022 Tutorial on Reproducibility in NLP with Jesse Dodge Jessica Zosa Forde Sasha Luccioni, PhD 🦋🌎✨🤗 Samarth Bhargav Robert Stojnic Koustuv Sinha ✈️ NeurIPS 🎉🥳

Accepted reports from MLRC 2021 is now published at ReScience journal! rescience.github.io/read/ To have a quick look at this years program, read our editorial here: rescience.github.io/bibliography/S… Cheers to our awesome AC's! Jessica Zosa Forde Jesse Dodge Sasha Luccioni, PhD 🦋🌎✨🤗 Robert Stojnic @sharath2269

Regarding Dalle and Imagen: These systems are *amazing*. However, I (selfishly) wish that all of that ML expertise and compute was focused on solving scientific problems, rather than generating panda art! Yes, it advances the field, but why not solve science simultaneously?

A seminar I attended taught by James Tompkin included dl.acm.org/doi/10.1145/12… - learning to read papers is an important skill to teach researchers!

It is really satisfying to see policy implementations following our work in National Academies study committee on "Reproducibility and Replicability in Science"—I'm proud of this service to the nation's scientific enterprise. #OpenScience #reproducibility doi.org/c5jp

I asked an ML person at a well-known lab recently whose moral responsibility it would be if their model suggested code that contained a serious bug and, due to automation bias, the user accepted the code, and the bug caused harm. They said the user's. This is why I'm terrified

#IJCAI2022 The winner of the IJCAI-22 John McCarthy Award is Michael L. Littman Michael Littman Brown University ⚡ Some Open Problems in Sequential Decision Making 🗓 29 July @ 13:45 – 15:15

NeurIPS.

PhD students presenting their findings to their supervisor:

🔥The ML Reproducibility Challenge is back! The 6th edition now expands to papers from 11 ML conferences: NeurIPS, ICML, ICLR, ACL, EMNLP, CVPR, ECCV, AAAI, IJCAI-ECAI, ACM FAccT & SIGIR. ML journals included this year: JMLR, TACL, & TMLR. Read more: paperswithcode.com/rc2022

a debugging manifesto