Fwd: MSR thesis committee & availability for thesis talk

Wed Apr 19 22:59:48 EDT 2023

Another big day for an Autonian: Ken Liu is defending his MSR thesis on
federated learning at 2pm!

Artur

---------- Forwarded message ---------
From: Ken Liu <kzliu at cmu.edu>
Date: Wed, Apr 19, 2023, 2:54 PM
Subject: Re: MSR thesis committee & availability for thesis talk
To: Steven Wu <zstevenwu at cmu.edu>, Virginia Smith <smithv at cmu.edu>,
Shengyuan Hu <shengyua at andrew.cmu.edu>, Elaine Shi <runting at gmail.com>,
Artur Dubrawski <awd at cs.cmu.edu>

Hi everyone,

Just a gentle reminder that my talk will happen 2pm ET tomorrow (location:
NSH 3305, zoom link
<https://cmu.zoom.us/j/97683261161?pwd=WTBGR3VSYXdvdmMvTkswa3dFWHNxdz09>)!
Please use the following forms for grading:

   - Speaking:
   https://www.ri.cmu.edu/app/uploads/2017/05/RI-MS-Speaking-Qualifier.pdf
   - Writing:
   https://www.ri.cmu.edu/app/uploads/2017/05/RI-MS-Writing-Qualifier.pdf

Thanks and hope to see you there!
Ken

On 5 Apr 2023, at 3:41 pm, Ken Liu <kzliu at cmu.edu> wrote:

Hi everyone,

Just quickly following up, we could use this zoom link in case you couldn’t
join in-person:

https://cmu.zoom.us/j/97683261161?pwd=WTBGR3VSYXdvdmMvTkswa3dFWHNxdz09

Thanks,
Ken

On 5 Apr 2023, at 3:25 pm, Ken Liu <kzliu at cmu.edu> wrote:

Hi everyone,

Thanks so much for providing your availabilities! The talk is now scheduled
to take place *April 20th from 2pm, at NSH 3305 (a seminar room on the
entry floor of the main RI building)*. We’ll have 2-3pm for the talk and
3-3:30pm to wrap up committee discussion (and you can leave early!).

*Re my thesis draft:* my MSR program prefers to have the thesis draft sent
to you 2 weeks before the talk, which is tomorrow (!). Unfortunately I
probably won’t have the draft ready by then, but just a gentle reminder
that the talk and the thesis will consists of the following items:

   - Our NeurIPS’22 paper on differential privacy in the context of
   cross-silo federated learning: https://arxiv.org/abs/2206.07902.
   - Our winning entry
   <https://drivendata.co/blog/federated-learning-pets-prize-winners-phases-2-3#puffle>
in
   the U.S.-U.K. Privacy-Enhancing Technologies (PETs) challenge
   <https://petsprizechallenges.com/> based on the above paper. Write-up /
   blogpost should also come soon.
   - Significant contribution to an ICLR’23 paper on private adaptive
   optimization (https://arxiv.org/abs/2212.00309).
   - Less significant contribution to an NeurIPS’22 FL workshop paper
   benchmark for personalization in federated learning (NeurIPS’22 FL workshop
   paper, https://arxiv.org/abs/2206.09262).
   - Results regarding an ongoing project interplay between gradient
   clipping and generalization techniques in private optimization. No write-up
   yet, but will be based on these notes/results
   <https://docs.google.com/document/d/1ncGQHDqtdMJwhN2PPDT97L8ohmcvqWCL_DUa8NEwwMI/edit?usp=sharing>
   .

I’ll be filling in the draft in the coming days/weeks at this link:
https://www.overleaf.com/read/dypzvpvjmkzk, and will also try to ping you
when the draft is readable. Thanks for your understanding!

Please let me know if you have any questions at all. Have a great day!

Best,
Ken

On 4 Apr 2023, at 9:25 pm, Steven Wu <zstevenwu at cmu.edu> wrote:

Looks like everyone could do Thursday 4/20 at 2pm. I could make that work
for me.

On Tue, Apr 4, 2023 at 3:14 PM Ken Liu <kzliu at cmu.edu> wrote:

> Hi everyone,
>
> I hope all has been well! Just another gentle reminder to please fill in
> your availabilities in the last two weeks of April for my thesis talk on
> this link! https://www.when2meet.com/?19437469-Y1XaA   (thanks so much if
> you’ve done so already!)
>
> *The talk needs not to be 90mins!* I checked with my program director and
> the usual allocation is to have 45 minute talk, 15 minute Q&A, up to 30
> minutes for committee discussion (which could be *much shorter* if you’re
> okay with my talk!). Filling in the form should take < 5mins.
>
> Thanks a lot!
> Ken
>
>
>
> On 2 Apr 2023, at 9:05 pm, Ken Liu <kzliu at cmu.edu> wrote:
>
> Hi everyone,
>
> I hope you’re doing well! A gentle reminder to please fill in your
> availabilities in the last two weeks of April on this link (
> https://www.when2meet.com/?19437469-Y1XaA) for my 90-min thesis talk.
> Thanks!
>
> Best,
> Ken
>
>
> On 31 Mar 2023, at 1:42 am, Ken Liu <kzliu at cmu.edu> wrote:
>
> Hi everyone,
>
> I hope you are all doing well, and thanks again for serving on my master’s
> (MSR
> <https://www.ri.cmu.edu/education/academic-programs/master-of-science-robotics/>)
> thesis committee! Just a quick reminder of the logistics & involvement:
>
>    - There will be *a 90-minute talk, on or before Apr 28 Friday*
>    - Filling in the forms listed here
>    <https://www.ri.cmu.edu/education/academic-programs/master-of-science-robotics/policies-and-forms-2/> (speaking/writing
>    qualifiers) during/after the talk
>
>
> My talk & thesis will involve the following:
>
>    - A project studying differential privacy in the context of cross-silo
>    federated learning (NeurIPS’22 paper, https://arxiv.org/abs/2206.07902
>    ).
>    - (New!) Our winning entry
>    <https://drivendata.co/blog/federated-learning-pets-prize-winners-phases-2-3#puffle>
>    in the U.S.-U.K. Privacy-Enhancing Technologies (PETs) challenge
>    <https://petsprizechallenges.com/>; we may expect more press releases
>    soon :).
>    - An ongoing project on the interplay between gradient clipping and
>    generalization techniques in private optimization.
>    - Contributions to other projects:
>       - private adaptive optimization (ICLR’23 paper,
>       https://arxiv.org/abs/2212.00309) and
>       - benchmark for personalization in federated learning (NeurIPS’22
>       FL workshop paper, https://arxiv.org/abs/2206.09262).
>
>
> *Would you please help me fill in your availabilities in the last two
> weeks of April on this link (https://www.when2meet.com/?19437469-Y1XaA
> <https://www.when2meet.com/?19437469-Y1XaA>)?* The last week might work
> the best for me!
>
> Thanks!
> Ken
>
>
>
>

-- 
Steven Wu
Assistant Professor
School of Computer Science
Carnegie Mellon University
zstevenwu.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20230419/a3f0f129/attachment.html>