Hai mai sognato di cavalcare al tramonto tra polvere e sangue, tra onore e vendetta, tra saloon affollati e duelli all’alba?#COLT2025 #COLTATaleofTwoCities #ColtATaleofTwoCities2025 #eventinerdottobre2025 #eventowesternItalia #giocodiruolodalvivo
www.corrierenerd.it/colt-a-tale-...

Universality of High-Dimensional Logistic Regression and a Novel CGMT under Dependence with Applications to Data Augmentation Over the last decade, a wave of research has characterized the exact asymptotic risk of many high-dimensional models in the proportional regime. Two foundational results have driven this progress: Gau...

Meanwhile, excited to be in #Lyon for #COLT2025, with a co-first author paper (arxiv.org/abs/2502.15752) with the amazing team -- Matthew M Mallory and our advisor Morgane Austern!

Keywords: Gaussian universality, dependent data, convex Gaussian min-max theorem, data augmentation!

COLT 2025 - Accepted Papers

Accepted papers for #COLT2025
learningtheory.org/colt2025/acc...

COLT 2025 - Call for Papers

Submit your favorite, itch-to-scratch, drive-you-mad, want-to-see-solved open problems in learning theory to #COLT2025!

More info: learningtheory.org/colt2025/cfp...

⏰ Deadline: June 6 AoE

Predictions and Uncertainty Workshop on Predictions and Uncertainty

📢 If you're interested in conformal prediction, algorithms w/predictions, robust stats & connections between them from a theory perspective, join us for a workshop at #COLT2025 in Lyon 🇫🇷 June 30!

Submit a poster description by May 25, more here:
vaidehi8913.github.io/predictions-...

How to safely discard features based on aggregate SHAP values SHAP is one of the most popular local feature-attribution methods. Given a function f and an input x, it quantifies each feature's contribution to f(x). Recently, SHAP has been increasingly used for g...

Ever aggregated SHAP values across sample points? Our #COLT2025 paper proves that this might be safe when your goal is to discard unimportant features - but only if you add one extra line of code that reshuffles your data! With Robi Bhattacharjee and Karolin Frohnapfel
arxiv.org/abs/2503.23111

Exciting news from our unit! 🎉
This year the ELLIS Unit Milan got 17 papers accepted at #ICML2025 and 3 at #COLT2025. Huge congratulations to all the authors!

More details about the papers will be shared closer to the conference dates.

Thank you to all of you who submitted a proposal, and a huge thanks to the workshop, tutorial, and community event organizers!

Please spread the word about these, and see you at #COLT2025!

Workshops at #COLT2025, CFPs and deadlines:

- Theory of #AI for Scientific Computing
tasc-workshop.github.io#cfp (⏰ May 16)

- Foundations of Post-training
fopt-workshop.github.io/cfp/ (⏰ May 19)

- Predictions and Uncertainty
vaidehi8913.github.io/predictions-... (⏰ May 25)

Community events and tutorials, list from the website

Workshops, list from the website

The tutorials, workshops, and community events for #COLT2025 have been announced!

Exciting topics, and impressive slate of speakers and events, on June 30! The workshops have calls for contributions (⏰ May 16, 19, and 25): check them out!
learningtheory.org/colt2025/ind...

Better Private Distribution Testing by Leveraging Unverified Auxiliary Data We extend the framework of augmented distribution testing (Aliakbarpour, Indyk, Rubinfeld, and Silwal, NeurIPS 2024) to the differentially private setting. This captures scenarios where a data analyst...

🎉 Clément Canonne's paper "Better Private Distribution Testing by Leveraging Unverified Auxiliary Data" has been accepted at #COLT2025! Check it out: arxiv.org/abs/2503.14709

🔗 More about COLT'25: learningtheory.org/colt2025/
👤 @ccanonne.github.io
#differential #privacy #testing #algorithms

Better Private Distribution Testing by Leveraging Unverified Auxiliary Data We extend the framework of augmented distribution testing (Aliakbarpour, Indyk, Rubinfeld, and Silwal, NeurIPS 2024) to the differentially private setting. This captures scenarios where a data analyst...

Our paper on "Better Private Distribution Testing by Leveraging Unverified Auxiliary Data" w/ Maryam Aliakbarpour, Arnav Burudgunte and Ronitt Rubinfeld will appear at #COLT2025!

Tackling the question: how to use historical data to assess properties of sensitive, new data?
arxiv.org/abs/2503.14709

Reminder: the deadline to propose a Tutorial, Workshop, or Community event at #COLT2025 in Lyon 🇫🇷 is tomorrow (April 18, AoE)!

learningtheory.org/colt2025/wtc...

Forgot to say: #COLT2025 is in Lyon (France 🇫🇷), June 30–July 4. (I am not on the PC or a main organizer, just tutorials, workshops, and community events chair 🪑 this year)

learningtheory.org/colt2025/wtc...

COLT 2025 - Workshops, Tutorials, and Community Events

📣 #COLT2025 will feature 3 types of satellite events: workshops, tutorials, and community events, the latter meant to bring together the learning community. Mentoring, affinity events, socials, you name it!

The call is here: learningtheory.org/colt2025/wtc..., submit your proposal by April 18, AoE!

i apparently just typed "uncertainty in the face of optimism" in a paper draft. maybe i need a bit of sleep, eh 🤣
#COLT2025