Florian had some convos with others on this topic, including Nicholas. We saw a topical conference with a call for position papers. We decided to write and submit a critical piece (self-critical of some of our previous works as well!) 3/n
@thegautamkamath
-
Privacy Challenges in Public Data for Machine Learning
By
–
This work started through some conversations with Florian (Summer 2022). We noticed that while public data was *really* helpful in private ML, weird things happened. An arms race of larger "public" datasets. Evaluating on datasets that didn't have much to do with privacy. Etc 2/n
-
Differential Privacy in Large-Scale AI Pretraining Wins ICML Award
By
–
"Position: Considerations for Differentially Private Learning with Large-Scale Public Pretraining," with @florian_tramer & Nicholas Carlini got an #ICML2024 best paper award! https://
x.com/thegautamkamat
h/status/1603383883126669312
… : the personal side of this research, emotional high & lows, & more 1/n -
Differential Privacy Foundations Published in Journal
By
–
Note that the original paper that introduced differential privacy was also published in JPC: https://
journalprivacyconfidentiality.org/index.php/jpc/
article/view/405
… -
Research Paper Journey: From arXiv to NeurIPS to Peer Review
By
–
We originally planned this paper as a cute note to live on arXiv. But people seemed interested, so we thought why not submit it to #NeurIPS2020. When we found out that the @uscensusbureau was going to use it, it needed a proper peer review. The only journal we considered was JPC.
-
Discrete Gaussian Differential Privacy Wins Caspar Bowden Award
By
–
Excited to share that "The Discrete Gaussian for Differential Privacy," joint work with Clément Canonne (@ccanonne_) and Thomas Steinke (
@shortstein
), was selected for the 2024 Caspar Bowden PET Award! Honoured for this work to be recognized. 1/2 -
Fine-Tuned Models Leak Private Pretraining Data Information
By
–
Nice talk by John Abascal on "TMI! Finetuned Models Leak Private Information from their Pretraining Data," a joint work with @the_stanleywu
, @AlinaMOprea
, and @thejonullman
. They show that fine-tuning large models can nonetheless leak info about their training data. -
UAI Conference Classification in AI Research
By
–
This is deflecting from Jon's question, but UAI is not a top ML conference. It's an AI conference.
-
PETS conference venue search leads to amusing misunderstanding at university
By
–
I was looking for the conference venue so I told someone at university "I'm looking for PETS" and they rubbed my head wtf?
-
ALT 2025 Conference in Milan: Submit Learning Theory Research
By
–
ALT 2025 is going to be in Milan, Italy , in February 2025. Po-Ling Loh and I are the PC co-chairs. Submit your best learning theory work!