Palak Jain

The recording of this week's talk, by Palak Jain (@thepalakjain.bsky.social), is now available online as well along with the slides:

"Enforcing Demographic Coherence: A Harms-Aware Framework for Reasoning about Private Data Release"
www.tcsplus.org/welcome/past...

Another TCS+ talk to look forward to: next week, Palak Jain (@thepalakjain.bsky.social) on a new data #privacy framework they introduce!
📋 forms.gle/ateNLcRo59H1... #TCSSky

The TCS+ calendar: Tom Gur on March 19 Or Zamir on April 9 Ryan Williams on April 23 Palak Jain on May 7

Bob* your calendar, as they say!

*Mark?

How a Columbia Student Fled to Canada After ICE Came Looking for Her Ranjani Srinivasan’s student visa was revoked by U.S. immigration authorities. That was just the start of her odyssey.

I am so angry about what is happening in the US, and to Columbia University students. The US government are entrapping students, lying about them, upending their whole lives.

"The department [of Homeland Security] did not provide any evidence for its allegations."

www.nytimes.com/2025/03/15/n...

Enforcing Demographic Coherence: A Harms Aware Framework for Reasoning about Private Data Release The technical literature about data privacy largely consists of two complementary approaches: formal definitions of conditions sufficient for privacy preservation and attacks that demonstrate privacy ...

It's been a hard few months. But today I was reading (finally) a paper I was looking forward to - also, heard lorikeets from the garden. There are good things in the world.
arxiv.org/abs/2502.02709

extra cute that I'm in this exact same situation 🫣

I use 'Souper Cubes' to freeze food 🥘

Dal and cooked rice also freeze well and are great for when you want something simple to eat.

Aside from the stuff that's already been mentioned, I find freezing a variety of curries is a great way to have a customisable meal for later, especially for difficult weeks. (Just warm up a curry of your choice and add veggies/paneer/fish that you have lying around)

Picture of attendees at workshop at encore institute

Thank you to EnCOREInstitute for having us and to all our amazing attendees!! It was such a great workshop thanks to the enthusiasm and engagement of everyone who attended encoredp.github.io

It was great to organise this with @ccanonne.bsky.social @amartyasanyal.bsky.social and Rachel Cummings

What a fantastic workshop! We need to do it again... An EnCORE?

This is a low-res sprite sheet of the pictures available

💡 Making slides, giving talks, and looking for illustrations for Alice, Bob, and the whole crew? Simón Oya (UBC) has drawn fresh, beautiful ones and releasing them under CC BY-SA 4.0 license: simonoya.com/drawings/

(discovered through the #EnCOREPrivacy25 workshop)

Theoretical CS Job Market 2024

It's tough to gain visibility as a young researcher, and it's job market season! Are you a theoretical computer science PhD/postdoc on the job market?

I don't have a crazy juge audience but I'll try to help a bit: fill this form, and I'll tweet your pitch and info!
docs.google.com/forms/d/e/1F...

I think you're exactly right. I can tell the poems apart precisely because the LLM poems read like they were written by a middle-schooler 😂

(8/8) Finally, I'm looking for a research internship this coming year and would love to set up a time to talk!

(7/8) We prove that this is the best achievable error bound that depends only on w, for a large range of values of w. When w is small, the error of our mechanism is similar to the polylogarithmic in T error in the insertion-only setting, bypassing the hardness in the turnstile model.

(6/8) We present an item-level differentially private mechanism that, for all turnstile streams with maximum flippancy w, continually outputs the number of distinct elements with an O(√w · poly log T ) additive error, without requiring prior knowledge of w.

(5/8) Specifically, the maximum flippancy is the largest number of times that the contribution of a single item to the distinct elements count changes over the course of the stream.

(4/8) We show that in the worst-case, every DP mechanism has additive error at least T^1/4! However, we can do much better in certain settings. -- We identify a parameter, maximum flippancy, that is low for natural data streams and for which we give tight parameterized error guarantees.

(3/8) With only insertions, existing algorithms have additive error just polylogarithmic in the length of the stream T. We uncover a much richer landscape in the turnstile model, even without considering memory restrictions.

(2/8) We consider the achievable error for differentially private continual release of a basic statistic—the number of distinct items—in a stream where items may be both inserted and deleted (the turnstile setting.)

(1/8) If you’re at NeurIPS '23, Iden Kalemaj and I will be presenting a poster (#1605) at 5pm today on joint work with Satchit Sivakumar, Adam Smith and Sofya Raskhodnikova on privately counting the number of distinct elements in turnstile streams.