Ian Waudby-Smith

@Yjg_oo There is some work out there on DP being used as a means to a statistical end (e.g. at the interface of privacy and robustness). In the above work, the end is "statistical inference for a population but with DP protections on the collected data".

@ianws0 In the past, the focus was on queries and private release of datasets. It appears that the stat community picked up DP for statistical procedures. Were there specific data attacking schemes that prompted this attention?Thank you!

At ICML at 11am Hawaiian time, I'll have a poster on differentially private confidence sets at board 801. Come hang out if interested! Paper: arxiv.org/abs/2202.08728

@edwardhkennedy Tiger is incredible

Very excited about this new paper by Tiger Zeng (tigerzhzeng.com) We study causal inference w/ high-dimensional discrete confounders We give new bias/variance results & minimax lower bounds, which characterize fundamental limits of causal inference in high dimensions

For the Stats crowd: unlike P-values, E-values (en.wikipedia.org/wiki/E-values) can be combined easily under arbitrary dependence. But can we say the same for their sequential analog, e-processes? Sadly the answer is *no*, motivating our recent work: arxiv.org/abs/2402.09698 (A thread)

@RdeHeide Wonderful news and well-deserved

@ChinmayaKausik Theorem 5.6 might be of interest. arxiv.org/abs/2401.15567…

Does anyone know if there is a matrix version of empirical Bernstein inequalities, as in this paper? I can use covering arguments to get one version, but that's not the "right" empirical equivalent of the matrix Bernstein inequality. arxiv.org/abs/0907.3740

@sp_monte_carlo A boring construction: X_1, X_1, X_1, ... Perhaps you want to impose that (X_t)_t has infinite variation (otherwise desideratum 4 leads to constant martingales).

@Apoorva__Lal Yussef Dayes! Here are some of my fav tracks. The first one takes a while to heat up but it's so worth the wait. open.spotify.com/track/0AGrnwML… open.spotify.com/track/2FCGnaXc… open.spotify.com/track/0gmjv0aT… open.spotify.com/track/4DPh5Nua…

been listening to a lot more drummers lately. here are some albums i like. recommendations? apoorvalal.lists.sh/drumAlbums

And giving an oral presentation in room 316 happening now.

@Apoorva__Lal @pedrohcgs Check out yabai (+skhd) next time you use a Mac! github.com/koekeishiya/ya…

@pedrohcgs Either setup works well if you use screen space. Default window snapping on windows (esp on w11) is good, linux has i3, and macos has ... nothing (last time i checked). I've installed rectangle whenever I've had to use a mac rectangleapp.com

Question for #EconTwitter. I am deciding between two 32'' monitors and an ultrawide 40+'' monitor for my work desktop setup. What are the pros and cons of each? What should I get?

Join us in person or online for our Discussion Meeting next Tuesday, 23 May. Your comments are keenly invited on the paper ‘Estimating means of bounded random variables by betting'. Register for the meeting at rss.org.uk/training-event…

@KrishnRamesh Congrats Krishn!!

our new home in Calgary 🙌🏽

Ramping up for this weekend’s #WiDSPittsburgh @CarnegieMellon events! Come join us for #datascience talks and networking with Keynote by Sheryl Skolnick, Strategic Advisor @UnitedHealthGrp @Optum! stat.cmu.edu/wids Registration is free! @WiDS_Worldwide @CMU_DietrichHSS

Shantanu Gupta, Emre Yolcu and Minji Yoon of @SCSatCMU and Ian Waudby-Smith of @CMU_DietrichHSS have been named 2023 @amazon Graduate Research Fellows. cmu.is/fellows-23

I am on the job market for faculty and research scientist positions. I develop statistical theory and algorithms to make data-based decisions that balance the need for robustness in high-stakes settings with strong performance in practice. Some highlights below: 1/n

@lakens @GregoryFaletto @LajeunesseLab This is true for simple settings (such as point nulls/alternatives in parametric problems) but in practice one might care about composite nulls/alternatives under nonparametric assumptions. For those settings, a bit more machinery is needed.

@GregoryFaletto @LajeunesseLab The sequential probability ratio test from the 50's also works.

There is tremendous risk in repeatedly analyzing data as it comes in… For example, in the absence of any effect, the p-values of repeated t-tests on a growing dataset will randomly walk between 1 and 0! Thanks to @lakens for inspiring this simulation! #RStats #dataviz

My advisor @edwardhkennedy and I have posted an article on continuous trt effects estimation: arxiv.org/pdf/2207.11825… We study nonparam models where the dose-response curve has its own smoothness, which may differ from that of the outcome regression or cond dens of trt given X.

Super excited to share new work "Comparing methods for statistical inference with model uncertainty" with @AdrianRaftery1 that just got published in PNAS (@PNASNews) ! #Statistics #DataScience #BMA (1/n) Paper: pnas.org/doi/10.1073/pn…

@Stat_Ron @CarnegieMellon @CMU_Stats Congratulations Ron! 🔥🎉

I find it hard to believe that 10 years ago I graduated high school to begin @CarnegieMellon - now I'm incredibly excited to share I will be an Assistant Teaching Professor @CMU_Stats! I owe everything to my amazing advisors, mentors, friends, & family for all of their support!