We obtain information-theoretically optimal list size and recovery error, and provide empirical comparison with prior methods.
link: https://t.co/hCRF4QOYK8
Joint with @rares_buhai, Stefan Tiegel, @alexxwolters, Gleb Novikov, @AmartyaSanyal, David Steurer, and @FannyYangETH.
Excited to present at #NeurIPS2024 our work on robust mixture learning!
How hard is mixture learning when (a lot of) outliers are present? We show that it's easier than it seems!
Join us at the poster session (Wed, 16:30 PT, West Ballroom A-D #5710).
Our method works in the presence of large outliers if mixture components are spherical Gaussians, or, more generally, have bounded k-th sub-Gaussian moments.
We propose a reduction from the robust mixture learning problem to a well-studied list-decodable mean estimation problem.
Open Postdoctoral position in Privacy (and unlearning) and Robustness in Machine Learning in University of Copenhagen to work with me!
If you want to spend a couple of years working on these exciting topics in beautiful Copenhagen, Apply here https://t.co/ImfAxifNG7
Happy to present today at #ICML (1pm, Hall C 4-9, Poster 1508) our poster on deep random features networks with structured weights (= with arbitrary covariances).
We obtain rigorous formulas for the generalization error and capture real networks trained with gradient descent.
We also prove Gaussian universality for networks with 2 hidden layers, extending collection of results about linearizations of neural networks.
Pass by if you are interested in how random matrix theory, Wiener chaos and Stein's method can be used in deep learning theory!
Very happy that this paper was accepted to #COLT2024 on the separation between private and non-private online learnability. Joint work with @dmitrievdaniil7 and Kristof Szabo.
There are still few open problems that remain to fully settle this question :)
The call for this position is now public.
This will be jointly supervised with Prof. Amir Yehudayoff at @DIKU_Institut , Prof. Varun Kanade @CompSciOxford (including a stay in Oxford for 3-6 months), and myself.
https://t.co/tMnTLeDl3R
RTs appreciated
2 concurrent papers, one by @inhaleopenair, @minilek, Lyu, Stemmer and Sarlós (https://t.co/Z6LSjFuXbp) & the other by us solve our open problem presented in #COLT2022. (See @GolowichNoah
& Livni's upper bound)
Interestingly, both have the same hard instance in the lower bound.
Excited about this new preprint with Daniil and @Kristo https://t.co/iBEubcFh69
We show a Ω(log T) lower bound for Differentially Private Online learning, even for finite Littlestone classes. This shows a separation between DP and non-DP online learning in mistake bound model.