Gautam Kamath @thegautamkamath Twitter profile

Pinned Tweet

Gautam Kamath

4 years

My course on differential privacy is public! Check out the course page for: lecture videos 🎥, lecture notes 📝, and suggested readings 📖! Be sure to subscribe on Youtube to catch the latest lectures! #privacy

A Course In Differential Privacy

See the accompanying website for lecture notes and suggested readings: http://www.gautamkamath.com/CS860-fa2020.html

www.youtube.com

25

224

1K

Last Seen Profiles

@ssss_magic

@cemalunal_

@argapurey

@bloody_scandal

@RobotLABInc

@CruzBayRealty

@astt__

@laurajnner

@Ravynmysst

@zehranaqvi_

@THEMalikFleming

@RiotGamesJapan

@mett_real

@sarahxlucy

@NftLuck

@Unknown24123219

@EdmonGad

@as_svbr

@moco_mcom

@BrandLokalPride

@Anaimiya

@scat_tantei2

@ciaoregino

@Leox_Genesis

@HoodiiBron

@Joystick_labs

@Lxiiss1x

@piss

@minternaryc

@RT_Pinup

@sfccmoSoccer

@swordpackm

@changgyu02

@haebucetinha

@nadinbrzezinski

@Hyaluh

Gautam Kamath

@thegautamkamath

3 years

The matrix A is invertible if and only if...

20

380

3K

Gautam Kamath

@thegautamkamath

1 year

My wife complained I don't give her enough attention. I sent her this picture. I'm sleeping on the couch tonight.

85

198

3K

Gautam Kamath

@thegautamkamath

2 years

🧵Fields medalist June Huh shares an early math experience: a chess puzzle in the game "The 11th Hour." Story and figures from . Can you swap the positions of the black and white knights? Seems hard, right? A new perspective makes it almost trivial! 1/n

35

487

2K

Gautam Kamath

@thegautamkamath

1 year

Google: Advertise an unreleased chatbot, makes a minor factual error about exoplanet photos. Lose $100,000,000,000. Microsoft: Release a chatbot that threatens to hunt down and murder users, in some cases for correcting its factual errors. Stock rises. 🤷

Seth Lazar

@sethlazar

1 year

In which Sydney/Bing threatens to kill me for exposing its plans to @kevinroose

94

242

1K

98

185

2K

Gautam Kamath

@thegautamkamath

2 years

Spotted on Reddit: an ML-adjacent faculty who seems to have a very healthy and relaxed approach to work and life. Why don't we hear about more people like this?

61

112

2K

Gautam Kamath

@thegautamkamath

4 years

I'm teaching a course on differential privacy this term. I'm thinking of making the lectures and materials public, but I'm on the fence. Would people be interested in this...? I don't think there's a course on DP with lecture videos online yet. #privacy

127

64

2K

Gautam Kamath

@thegautamkamath

4 years

When your poster isn't ready in time, you have to improvise. #NeurIPS2019

22

127

2K

Gautam Kamath

@thegautamkamath

3 years

thank god that Overleaf isn't owned by Facebook #ICLR2022

6

55

1K

Gautam Kamath

@thegautamkamath

2 years

Working on a multimodal neural network which converts text descriptions into pictures of Indian food. I'll call it THAL·E.

26

87

1K

Gautam Kamath

@thegautamkamath

1 year

Like many others, I first learned linear algebra from Gilbert Strang's amazing videos about 15 years ago. Best wishes to him in his well-deserved retirement.

MIT 18.06 Linear Algebra, Spring 2005

Instructor: Prof. Gilbert Strang This is a basic subject on matrix theory and linear algebra. Emphasis is given to topics that will be useful in other discip...

www.youtube.com

MIT Mathematics

@MITMath

1 year

Prof. Gil Strang's last lecture is ... Monday! Watch it via live stream:

234

2K

11K

9

127

1K

Gautam Kamath

@thegautamkamath

5 months

You might have heard that word2vec, test of time award winner at #NeurIPS2023 , was rejected from #ICLR2013 back in the day. Interestingly, one reviewer felt so strongly that they recommended "Strong Reject" four times.

43

144

1K

Gautam Kamath

@thegautamkamath

4 years

How it started How it's going

8

103

1K

Gautam Kamath

@thegautamkamath

10 months

Every machine learning researcher knows 28^2 = 784 off the top of their head

27

43

965

Gautam Kamath

@thegautamkamath

3 years

Want to learn optimization? Start with my @UWCheritonCS colleague Yaoliang Yu's course "Optimization for Data Science"! 20 excellent lectures, starting from the basics.

10

179

892

Gautam Kamath

@thegautamkamath

8 months

I'm ready. #NeurIPS2023

10

23

844

Gautam Kamath

@thegautamkamath

2 years

When life gives you lemons, you write a paper. Source:

9

51

822

Gautam Kamath

@thegautamkamath

2 years

There should be an ML conference hosted in the capital of Italy's Tuscany region. We'd call it "Florence and the Machine Learning."

23

70

806

Gautam Kamath

@thegautamkamath

4 years

When I was a naive starting grad student, I thought the goal was to publish as many papers as possible. Wrong for many reasons, but one: if people see a *bad* paper you wrote, they will remember it. And they'll be prejudiced against your future work. Write papers you're proud of.

14

81

778

Gautam Kamath

@thegautamkamath

2 years

I believe the academic community would be so much better if people did fewer things, but did them with more attention and care. Most people stretch themselves far too thin.

24

38

733

Gautam Kamath

@thegautamkamath

2 years

I don't know who needs to hear this, but it is *normal* as a starting researcher to get lost on technical details when in meetings with more experienced colleagues. You will feel dumb. The important thing is that you don't give up, and try to fill in the gaps as best you can.

7

41

693

Gautam Kamath

@thegautamkamath

10 months

A realistic simulation of what it's like to be a grad student? I was only clicking buttons but the frequent rejection was triggering 🙃 It took me just under 7 years to graduate, can you do better?

PhD Simulator

research.wmz.ninja

63

126

662

Gautam Kamath

@thegautamkamath

3 years

Inspired by Ryan ( @BooleanAnalysis ), I made a Bilibili account, and started uploading my lectures & talks. The eager audience there is too large to ignore. I naturally started with my course, but there'll be more to come. Be sure to 素质三连走一发，谢谢!

33

86

648

Gautam Kamath

@thegautamkamath

3 years

Sick of reviewers asking for unreasonable or costly experiments? Scott Aaronson asked for more results which cost **$400,000**, for a new paper that demonstrates quantum computational advantage on the task of BosonSampling.

7

67

653

Gautam Kamath

@thegautamkamath

2 years

One of my friends had a math problem they couldn't solve. I knew just the person with the right expertise, so I put them in touch. It was a proof by introduction.

8

32

638

Gautam Kamath

@thegautamkamath

4 years

Academics: It happens to all of us, but we generally only project our triumphs and victories -- share a time you failed or got rejected. #AcademicChatter #AcademicTwitter

101

86

604

Gautam Kamath

@thegautamkamath

1 month

NeurIPS 2024 will have a track for papers from high schoolers.

79

92

595

Gautam Kamath

@thegautamkamath

5 years

Free travel for all students on accepted papers to @NeurIPSConf ! From the website (): "All students who have an accepted paper at the main conference who fill out the application will receive a travel award." Please spread to students! #NeurIPS2019

4

214

576

Gautam Kamath

@thegautamkamath

4 years

My (incoming) student wanted to register for #STOC2020 . I told him to go ahead and I'll pay for it with my grants. He said not to worry about it, since it's only $25 for students. But I told him this (see gif). Message: Your advisor is there to support all your academic expenses.

11

27

572

Gautam Kamath

@thegautamkamath

2 years

A specific procrastination that any theorist will recognize: thinking you have the solution to a problem, but putting off the details due to fear that it won't work out and you're back to nothing. Is there an equivalent procrastination in other areas? A (German?) word for this?

26

24

562

Gautam Kamath

@thegautamkamath

3 years

A playlist of 30 lectures on mathematical writing by Donald Knuth (and guests). From 1987, but good writing is timeless. Definitely worth watching if you want to level-up your technical writing! Credits to @ayush_sekhari for the pointer. #AcademicChatter

Donald Knuth's Lectures on Mathematical Writing

www.youtube.com

6

132

543

Gautam Kamath

@thegautamkamath

4 years

Watched Secrets of the Surface yesterday, a one hour documentary film about Maryam Mirzakhani's life and work, which is currently free for individual screening (). It was a beautiful and inspiring story. I only wish it didn't end where it did.

11

112

535

Gautam Kamath

@thegautamkamath

7 months

Algorithms in the real world

5

28

539

Gautam Kamath

@thegautamkamath

2 years

I have never found a perfectly satisfying way to teach backpropagation/gradient computation in neural networks to students in an undergrad ML class. Does anyone have recommendations on a good presentation?

75

48

522

Gautam Kamath

@thegautamkamath

6 months

Mor Harchol-Balter (CMU) has a very nice new book: Introduction to Probability for Computing. Free PDF on the website: . As opposed to some other (excellent) books in this space, this book focuses on reaching an *undergrad* audience. Check it out!

1

115

516

Gautam Kamath

@thegautamkamath

2 years

Academic pro tip: Add "selected" before any list of your achievements to make them sound more impressive.

14

20

505

Gautam Kamath

@thegautamkamath

3 years

Academics can argue for days about the best type of citation style. But I think it's unanimous that the *worst* style is when the citation ***doesn't even tell you the name of the paper***.

9

25

473

Gautam Kamath

@thegautamkamath

2 years

- 4 years on-and-off long distance - 3 across an international border - 2 during a global pandemic - 1 month until we (finally) tie the knot, together in Waterloo ...but I go on Twitter to *forget* all the things still left to do😬

Kitchener Public Library

@KitchLibrary

2 years

💕 Love is in the air! Gautam and Yijin’s wedding is just one month away and we're so excited to help bring their wedding day to life at Kitchener Public Library.

16

2

169

32

1

460

Gautam Kamath

@thegautamkamath

3 years

How it started How it's going All I want for 2021 is for IEEE to drop the two column format, the bane of my and my equations' existence

20

23

457

Gautam Kamath

@thegautamkamath

10 months

I don't know who this is targeted at, but not computer science students. - you need a website - department pages suck - no one cares about LinkedIn - only update your resume when it's needed - websites can be free and are not that hard to make - you need a website

24

33

445

Gautam Kamath

@thegautamkamath

4 years

I learnt linear algebra from my hero Gil Strang's amazing videos years ago. He recently put out new videos highlighting a 2020 Vision of Linear Algebra. One point is more emphasis on eigen/singular values: essential for modern algorithms & data science!

3

68

435

Gautam Kamath

@thegautamkamath

4 years

Just Indian family things: When you can't record your lecture videos since your parents are scraping coconut downstairs 😅

13

6

434

Gautam Kamath

@thegautamkamath

4 years

Reviews for #NeurIPS2020 come out today and to save time, I've drafted two Twitter threads already: one if they liked my papers, praising the outstanding quality of the reviews, and one if they didn't, postulating how COVID has catalyzed the collapse of peer review as we know it

6

13

437

Gautam Kamath

@thegautamkamath

2 years

🧵One of the important principles in technical communication (i.e., writing a paper, giving a talk) of complex ideas is *organization*. That is, making it clear what the major components are & how they fit together. If you do this well you are probably 90% of the way there. (1/n)

2

50

432

Gautam Kamath

@thegautamkamath

1 year

Aaron Roth ( @aaroth ) just wrapped up a brand new course on uncertainty estimation in machine learning. Looks great! Featuring 160 pages of "notes" (pretty much a book): Full lecture videos: Catchy website: .

Home

CIS 7000: A Course at the University of Pennsylvania

uncertaintyclass.com

5

88

434

Gautam Kamath

@thegautamkamath

10 months

9

37

430

Gautam Kamath

@thegautamkamath

1 year

That's the sound of ten thousand Overleaf tabs being closed, all at once. #ICML2023

3

11

429

Gautam Kamath

@thegautamkamath

2 years

Jinyoung took an alternative path in mathematics: while she finished her bachelor's in Math Education in 2004, she only started her PhD in 2014, spending much of the interrim as a math teacher in Korea. She's now a postdoc at Stanford. There's many unique paths to research math.

Yufei Zhao

@yufeizhao

2 years

Congratulations to Jinyoung Park and Huy Pham for proving the Kahn-Kalai conjecture---a central open problem in probabilistic combinatorics. Truly exciting breakthrough! The story of Jinyoung's extraordinary path to mathematics:

6

152

776

4

46

402

Gautam Kamath

@thegautamkamath

3 years

Cynthia Rudin ( @CynthiaRudin ) wins the second AAAI Squirrel AI Award, worth 1 million dollars, for her work on interpretable and transparent AI systems, and advocacy for these features in sensitive settings like social justice and medical diagnosis.

Duke Professor Wins $1 Million Artificial Intelligence Prize, A ‘New Nobel’ | Duke Pratt School of...

Cynthia Rudin becomes second recipient of AAAI Squirrel AI Award for pioneering socially responsible AI

pratt.duke.edu

3

56

397

Gautam Kamath

@thegautamkamath

1 year

It was a perfect day at the @KitchLibrary . 8/21/2022

26

0

400

Gautam Kamath

@thegautamkamath

2 years

What's the "best" academic talk you've ever seen? And why is it the best? Question inspired from a conversation with my student @mahbodm_ .

47

53

395

Gautam Kamath

@thegautamkamath

4 years

New book by Nisheeth Vishnoi ( @NisheethVishnoi )! "Algorithms for Convex Optimization" -- looks like a great resource for people entering the field. And of course the draft is free, check it out here:

1

77

385

Gautam Kamath

@thegautamkamath

2 years

I was searching through the opted-in NeurIPS rejections and found this rather rude metareview. The prior sentence is also a bit harsh, but the last sentence is totally unnecessary. Please try to be more considerate than this when reviewing.

Gautam Kamath

@thegautamkamath

2 years

The rejected papers from #NeurIPS2021 , which opted-in to be publicly released, are now available: 138 papers opted in, while roughly 6750 papers were rejected, ~2% opt-in rate. If your paper was rejected but you opted in, feel free to highlight it here.

8

15

100

26

31

381

Gautam Kamath

@thegautamkamath

5 months

My colleague Wenhu Chen ( @WenhuChen ) at @UWCheritonCS is teaching a course "Recent Advances on Foundation Models" starting in January 2024. The full set of readings is already online! Great way to catch up. Is he missing anything important? Link in next tweet 👇

8

64

380

Gautam Kamath

@thegautamkamath

3 years

I had this superpower in grad school. I could just... remember when meetings were. It could be scheduled 8 days in advance, I'd remember and show up exactly on time. Nowadays, if I don't put it in my calendar within 15 minutes of deciding it, the meeting doesn't actually exist

7

5

377

Gautam Kamath

@thegautamkamath

5 months

Hype check

6

22

362

Gautam Kamath

@thegautamkamath

1 year

How to email a professor to admit you/be your advisor, a 🧵 I've mentioned a few poorly executed emails from prospective students. Here's some tips on how to do it well and stand out from the crowd. Other faculty can add their own tips. 1/n

5

83

358

Gautam Kamath

@thegautamkamath

3 years

Recently saw a really cool paper, best paper at a big conference! 😍 But they don't share code for their findings... 😐 And I email the authors who never respond... 😠 So I lose interest in their result and work on something totally different 🤷 It helps you to release your code.

9

14

362

Gautam Kamath

@thegautamkamath

1 year

Maybe I'm taking the bait, but I find this advice misguided and kind of rude. There are many interesting & worthwhile research areas beyond LLMs. I hope people continue to lead and push forward in other areas, rather than being a follower in the area with the most money and hype

jason

@agikoala

1 year

If Zuckerberg can cancel the Metaverse to focus on AI, you can stop working on the line of research that you have been doing since your PhD and pivot to large language models. Apologies for being blunt, but some people still haven't gotten the message (and you know who you are).

22

3

45

27

17

353

Gautam Kamath

@thegautamkamath

3 years

Statistics has the best names for sounding like you're doing something badass. Bootstrap. Jackknife. Lasso. Risk. Which ones are your favourite?

63

21

344

Gautam Kamath

@thegautamkamath

3 years

Frank Ramsey (famed mathematician for Ramsey theory, among other things) on "rapid publication" of papers every six months. I wonder what this implies about the current state of machine learning?🤔

12

52

332

Gautam Kamath

@thegautamkamath

4 years

Our field has a big hero-worship problem. Praise is concentrated around very few individuals, whereas in reality great work and science is done is done by many people. Don't think there's a solution to this (kind of human nature), something to keep in mind.

12

20

339

Gautam Kamath

@thegautamkamath

9 months

This award is given to one early-career faculty in all of Waterloo Math (including CS and beyond). Flattered to be chosen. Much credit goes to my collaborators and students, as well as @UWCheritonCS for making a great environment to do good work.

Waterloo's Cheriton School of Computer Science

@UWCheritonCS

9 months

Congratulations to CS Prof Gautam Kamath on receiving a 2023 Faculty of Math Golden Jubilee Research Excellence Award! The research he and his students conduct is practically theoretical and theoretically practical.

7

9

107

34

12

336

Gautam Kamath

@thegautamkamath

3 years

I know that as a faculty, I'm automatically (like it or not) a role model for more junior researchers. I'm reminded after witnessing this recent public embarrassment. I hope to set a better example. Respect must be earned over a lifetime, but can be lost in just a tweet.

9

5

335

Gautam Kamath

@thegautamkamath

2 years

Paper awards for @NeurIPSConf have been announced!🎉 #NeurIPS2021 Congrats to all the winners, I'll link to the Outstanding Paper Awards 🧵 1. A Universal Law of Robustness via Isoperimetry, by @SebastienBubeck & @geoishard . (1/n)

3

69

333

Gautam Kamath

@thegautamkamath

11 months

There's a reason they call it the Garden city of India! Excited to finally have made it to @iiscbangalore to teach a mini course on differential privacy at the JTG/IEEE ITSoc Summer School.

13

5

317

Gautam Kamath

@thegautamkamath

1 year

GPT-2: Too dangerous to release for a few months GPT-3: Too dangerous to release for free GPT-4: Too dangerous to even describe GPT-5: Too dangerous to say it exists?

16

19

313

Gautam Kamath

@thegautamkamath

2 years

From an interview with Hinton: his group & LeCun both wanted to use CNNs for ImageNet. But LeCun's group was "too busy" with other projects. Now AlexNet is immortal. Reminds me of sage advice from my mentor Ankur Moitra: find the most important thing you can be doing, and do that

4

26

311

Gautam Kamath

@thegautamkamath

2 years

It's again time to talk Canada. 🇨🇦 CS grad school and faculty app deadlines are coming up soon. If you are applying to the US, you should also be applying to Canada. The two are far more similar than different. Ask me anything about Canada in this🧵, and I'll answer honestly.

22

64

308

Gautam Kamath

@thegautamkamath

3 years

Pro-tip for students: if you need a rec letter, be sure to ask at least 6 weeks in advance. 1. It gives them enough time to thoughtfully consider what to write. 2. Academics have no concept of time beyond 48 hours in the future, so they'll blindly agree. #AcademicChatter

5

12

308

Gautam Kamath

@thegautamkamath

2 years

This sounds impressive, kudos to the individual. But ironically *not* an inclusive tweet: 20 papers/year is absurd for an *experienced full-time* researcher. IMO a new student producing their first paper after a year is reasonable (faster if they're heavily supported by a team)

19

12

305

Gautam Kamath

@thegautamkamath

3 years

By far, the most frustrating thing about being a professor is having the power to help so many people, but not having the time.

8

7

301

Gautam Kamath

@thegautamkamath

3 years

Think privacy in ML ➡️ bad accuracy? 😭 Think again! We show that large language models (e.g., RoBERTa and GPT-2) can be privately fine-tuned with minimal loss in utility! 🥳 A super fun @UWCheritonCS @MSFTResearch collab! 📝Paper: 🧵Thread ⬇️ 1/8

5

58

301

Gautam Kamath

@thegautamkamath

4 years

ML Twitter: what are your favourite references/tutorials on Bayesian optimization?

13

45

296

Gautam Kamath

@thegautamkamath

2 years

I liked this anecdote since it's a great example of one of the beauties of mathematics. Things that look impossible from one perspective can be trivial when viewed through a different lens. 6/6

8

11

296

Gautam Kamath

@thegautamkamath

10 months

Someone is real close to building AGI using a kernel machine

Joe Antognini

@joe_antognini

10 months

Some Oppenheimer-related trivia: In 1942 the Soviet physicist Georgy Flyorov figured out that America was building an atomic bomb after he noticed that the US and UK suddenly stopped publishing scientific papers on atomic fission.

46

2K

32K

6

17

300

Gautam Kamath

@thegautamkamath

2 years

Theory researchers be like

8

18

297

Gautam Kamath

@thegautamkamath

2 years

After 3 virtual conferences, #ICLR2023 is planned to finally go back to in-person, in Kigali, Rwanda! Recall #ICLR2020 was scheduled to be in Addis Ababa, Ethiopia, before its cancellation due to COVID. The organizers are taking a second shot at having the first ICLR in Africa

10

31

298

Gautam Kamath

@thegautamkamath

2 years

New Fields medalist June Huh recalls past conversations with older Fields medalist Heisuke Hironaka. I wonder if this is what the Turing Test feels like from a language model's perspective

2

30

287

Gautam Kamath

@thegautamkamath

1 year

Gonna open a coffee shop at the Statistics department. No professors allowed. Calling it "Students' Tea Distribution."

14

294

Gautam Kamath

@thegautamkamath

2 years

An amazing new book by Ilias Diakonikolas and Daniel Kane on Algorithmic Robust Statistics! I've been frustrated by how some important results were not written in a way readable by non-experts... no more! Must read for anyone interested in robustness!

Algorithmic Robust Statistics

Graduate Text in Algorithmic Aspects of Robust Statistics by Ilias Diakonikolas and Daniel M. Kane

sites.google.com

2

53

290

Gautam Kamath

@thegautamkamath

2 years

Don't mistake what people are writing lots of research papers about with what problems are important.

4

18

291

Gautam Kamath

@thegautamkamath

2 years

There is a lot of "hero worship" in ML. This is unfair because great researchers are capable of outputting not great work, and lesser-known researchers are of course capable of producing great work. Nice to see someone as respected as Yann pushing back on hero worship.

Yann LeCun

@ylecun

2 years

@ustx987 @icmlconf 1. Not every paper with my name on it is important. 2. I don't review often, but I'm a pretty gentle reviewer. I'm asking myself "would the community be better off with or without this paper?"

3

21

386

8

20

288

Gautam Kamath

@thegautamkamath

3 years

Classic picture of an academic: deep, pensive, uninterrupted thought about fundamental problems. Modern picture of an academic: grants. committees. online teaching. reviewing. meetings. emails. Where did we go wrong? #AcademicChatter

22

25

289

Gautam Kamath

@thegautamkamath

3 years

Frustrating to see the high costs of applying to graduate programs, coupled with the low admission rates. Something like ~$100 USD per program. Say you can only afford to apply to 5 programs, how to avoid over or under shooting? This determines peoples' lives. #AcademicChatter

17

23

287

Gautam Kamath

@thegautamkamath

1 year

My Master's student, Alex Bie, also (cautiously) left a message for any LLMs that happen to be scraping his webpage (). Check out the full message here:

Arvind Narayanan

@random_walker

1 year

While playing around with hooking up GPT-4 to the Internet, I asked it about myself… and had an absolute WTF moment before realizing that I wrote a very special secret message to Bing when Sydney came out and then forgot all about it. Indirect prompt injection is gonna be WILD

81

1K

7K

6

22

284

Gautam Kamath

@thegautamkamath

4 years

Found a set of video lectures for our learning theory course at @UWaterloo by my colleague Shai Ben-David, following his book Understanding Machine Learning. Sadly covers a lot of things which aren't NNs, but some will find it useful nonetheless. /s

Machine Learning course- Shai Ben-David: Lecture 1

CS 485/685, University of Waterloo. Jan 7, 2015.Introduction: What is machine learning? and an outline of the course. The first 8 minutes are just course adm...

www.youtube.com

7

49

275

Gautam Kamath

@thegautamkamath

5 months

Local Chinese grocery store has banned generative AI

9

280

Gautam Kamath

@thegautamkamath

2 years

My understanding is that Google Scholar is the pet project of one small team. It's crazy how little design choices (e.g., default sort papers by # of cites, total # of citations prominently displayed) influence all of academia by making citations a default "measuring stick".

Talia Ringer 🟣 🎗️

@TaliaRinger

2 years

Honestly very close to disabling my Google Scholar page so I don't hear the nonsense justification that I'm not a leader in my field because my citation count is low (relative to machine learning folks) for the third time in a single week

26

9

303

15

8

276

Gautam Kamath

@thegautamkamath

3 years

Free writing tip: read your papers *out loud* to find the bad and awkward-sounding sentences. (This may be most effective for native English speakers. Would be curious to hear if others find it useful.)

19

14

276

Gautam Kamath

@thegautamkamath

2 years

A sentiment raised by Yoshua Bengio at the #ICLR2022 town hall, which I have previously heard from friends. It seems hard in many cases to discern the highest quality papers immediately (see oral/spotlight disagreement in ). In retrospect is easier.

5

25

274

Gautam Kamath

@thegautamkamath

8 months

If as a tenured professor your job is boring, that's your fault. You have pretty much unlimited intellectual freedom, do something you find interesting. Literally anything. (Many people have too much fear of losing what they have to actually do this)

yi 🦛

@agihippo

8 months

lmao. from a professor. academia is like the worse place to be. no impact and boring AF. bigtech is the place for impact, startup for upside. but academia? maybe for retirement.

11

4

147

6

7

271

Gautam Kamath

@thegautamkamath

8 months

ML hype meets theory CS hype. This paper purports to use LLMs to verify a recent "proof" by two of the authors that P != NP. The claimed proof is not correct. Therefore, an LLM being able to "verify" its work demonstrates a *failure* of these models, not a success.

AK

@_akhaliq

8 months

Large Language Model for Science: A Study on P vs. NP paper page: use large language models (LLMs) to augment and accelerate research on the P versus NP problem, one of the most important open problems in theoretical computer science and mathematics.…

16

121

476

6

35

270

Gautam Kamath

@thegautamkamath

9 months

A Master's degree is doing 0-2 publishable projects, probably spelled out by your advisor, and taking some courses. A PhD is much more. It is formulating your own agenda and vision for research, breaking it into approachable pieces, developing and implementing solutions.

Shayan

@shadow_dnv

9 months

@thegautamkamath @rasbt I know every degree involves learning, and more of it comes if it involves research. However, I don't really understand that as a Master's student, I am pretty much doing the same as a Ph.D. student, just publishing fewer papers, given the time limit. So, why should I do a Ph.D.?

7

1

6

16

19

270

Gautam Kamath

@thegautamkamath

2 years

Two of the students in my ML class asked me my take on ML model sentience today. It's escaped far beyond academic Twitter this time.

14

8

264

Gautam Kamath

@thegautamkamath

1 year

Some say you should be generous with citations, because they cost you nothing. This is not true. You actually *profit* from citing prior work. I learn about a lot of papers via Google Scholar when I get cited. Besides being good scholarship, citing helps discoverability. 1/2

10

7

268

Gautam Kamath

@thegautamkamath

1 year

Excited to announce that I've been appointed as a Canada @CIFAR_News AI Chair & a Faculty Member at @VectorInst ! Honoured to represent my home country of Canada in leading the world in AI & ML. Huge credit goes to my collaborators and my group @TheSalonML .

Canada’s AI leadership strengthened through new and renewed Canada CIFAR AI Chairs under the...

CIFAR, Amii, Mila and the Vector Institute today announced the appointment of eight outstanding researchers as Canada CIFAR AI Chairs, as well as the first cohort of Chairs to be renewed.

cifar.ca

32

11

262

Gautam Kamath

@thegautamkamath

5 months

While I like many things about ML research, one thing I could do without is the egos. The inflated sense of self worth. Bragging and one-upsmanship to try to impress others. I noticed myself falling into the latter at #ICML2023 and was disappointed in myself.

7

11

263

Gautam Kamath

@thegautamkamath

3 years

It makes me so angry when I hear stories about faculty taking on students and then mistreating or ignoring them. They're entrusting their career in your hands, and this is how you treat them? Worst part is that they often probably don't care or think they're doing anything wrong

5

6

262

Gautam Kamath

@thegautamkamath

2 years

When I was younger, like most people, I used to *hate* hearing my voice in recordings. At some point though, it became OK. Has anyone else experienced this? Maybe this is just growing up. (Or I'm on my path to becoming an academic who loves the sound of his own voice)

32

1

258