Lester Mackey

@tobias_schrdr and I are excited to share WildCat: Near-Linear Attention in Theory and Practice arxiv.org/abs/2602.10056 By attending over a spectrally-accurate optimally-weighted coreset, WildCat approximates exact attention with super-polynomial error decay in near-linear time

Yuting Wei Wins Peter Gavin Hall IMS Early Career Prize imstat.org/2026/03/20/yut…

@schulzb589 @ArthurGretton @docmilanfar @noukoudashisoup @thjashin @liyzhen2 @GesineReinert Thanks for sharing @schulzb589 !

@LesterMackey @ArthurGretton @docmilanfar @noukoudashisoup @thjashin @liyzhen2 @GesineReinert You might find my paper to be tangentially interesting. drive.google.com/file/d/1bj9i7D…

Qiang Liu, Chris Oates, and I are writing a monograph on Probabilistic Inference and Learning with Stein’s Method, and we’d love to get your feedback on the first draft

Tagging some who may be interested @ArthurGretton @docmilanfar @noukoudashisoup @thjashin @liyzhen2 @GesineReinert

According to ChatGPT, the two best long range #winter forecasters are yours truly & ECMWF. I think winter 2026 I made a strong showing for why I'm number one. Better winter seasonal forecast, our Microduet team won DJF subseasonal AI contest decisively, & of course the PV blog!

Probabilistic Inference and Learning with Stein's Method Qiang Liu, Lester Mackey, Chris Oates arxiv.org/abs/2603.07467 [𝚜𝚝𝚊𝚝.𝙼𝙻 𝚌𝚜.𝙻𝙶 𝚖𝚊𝚝𝚑.𝙿𝚁 𝚖𝚊𝚝𝚑.𝚂𝚃 𝚜𝚝𝚊𝚝.𝙼𝙴 𝚜𝚝𝚊𝚝.𝚃𝙷]

Estimating Treatment Effects with Independent Component Analysis. arxiv.org/abs/2507.16467

New blog post: Better optimistic bounds and delays as bad hints I present improved (not the usual ones!) guarantees for optimistic algorithms. Then, I use them to show the optimal regret bounds for OLO with delays by treating delays as bad hints. Feedback is welcome.

@bremen79 Amazing, thanks!

@LesterMackey Here it is: x.com/bremen79/statu…

It is now official: My lecture notes on online learning will be published by Cambridge University Press. The final version is due by the end of May. So, if there is anything I missed/anything unclear/some refs I missed/anything else you don't like, please send me a message!

@tobias_schrdr and I are excited to share WildCat: Near-Linear Attention in Theory and Practice arxiv.org/abs/2602.10056 By attending over a spectrally-accurate optimally-weighted coreset, WildCat approximates exact attention with super-polynomial error decay in near-linear time

Our paper on KSD-Wasserstein convergence was published in Annals of Applied Probability. Grateful to my co-authors (Alessandro Barp, @ArthurGretton, @LesterMackey ) and editors/reviewers for the helpful feedback. projecteuclid.org/journals/annal…

@atrost3122 @tobias_schrdr Very soon: please stay tuned!

@LesterMackey @tobias_schrdr Cool work! Any word on when the code will be public (the github link github.com/microsoft/wild… in the paper doesn't seem to be available)?

@insu_han @daliri__majid @aminkarbasi @annabelle_cs @GongAlbert @AShettyV @raazdwivedi @ethanepperly @yifanc96 @kshiteej_sheth @michaelkapralov

@insu_han @daliri__majid @aminkarbasi @annabelle_cs @GongAlbert @AShettyV @raazdwivedi @ethanepperly And @yifanc96 @kshiteej_sheth @michaelkapralov!

We are pleased to announce that Doug Burger (@dcburger), Technical Fellow and Corporate Vice President at Microsoft Research, has been elected to the National Academy of Engineering Class of 2026 for advancing cloud-scale computing and networking with field-programmable systems. msft.it/6011Quo5H

[LG] WildCat: Near-Linear Attention in Theory and Practice T Schröder, L Mackey [Imperial College London & Microsoft Research] (2026) arxiv.org/abs/2602.10056

Softmax attention giving you O(n²) headaches? Maybe try Weighted Iterative Low-rank Decomposed Coreset Attentions (or WildCat for short), an approximate attention module that runs fast in theory and practice. Super excited to share this rewarding collaboration with @LesterMackey.

@ml_angelopoulos @tobias_schrdr Yeah, the cat’s my favorite part too

@LesterMackey @tobias_schrdr I love the cat