Clément Canonne
Senior Lecturer #USydCompSci at the University of Sydney. Postdocs IBM Research and Stanford; PhD at Columbia. Converts ☕ into puns: sometimes theorems. He/him.
- Finally started my Victorian-era whodunnit novel! Any feedback welcome. "It was the best of crimes. It was the worst of crimes. It was
- [Not loaded yet]
- Not sure the "Basic" (free) version of Gradescope allows that, but worth a look? www.gradescope.com
- Getting ready to celebrate our graduating students at the School of Computer Science's Award Ceremony! @sydneycompsci.bsky.social
- Congratulations to Shuei Sakaguchi, recipient of the Allan Bromley Prize – awarded to the best Honours thesis! Shuei is now a PhD student in the Sydney Algorithms and Computing Theory (SACT) group!
- Roses are red Violets are blue Overfull \hbox (53.04803pt too wide) in paragraph at lines 136--163 And so are you
- [Not loaded yet]
- [Not loaded yet]
- That's a different question. You are expected to do research and publish, you have some say about *where*.
- No, the salary from *the university*. Research and service to the research community are part of the job we are paid for.
- (whether it's paid or incentivized enough, esp. the latter, is a different question. But it's in the job description.)
- This... This seems to be true. Did you know? WHY DID NO ONE TELL ME? www.abc.net.au/news/science...
- From some light testing, it actually seems like adding -x, for any x, turns off the "AI overview". It might simply be that Google (for now) doesn't trust its AI responses to handle exclusions.
- Ah... So it may not be a hidden feature, just temporary relief...
- Highly recommend Tom Gur (@tomgur.bsky.social)'s TCS+ talk on the Zero-Knowledge PCP theorem his coauthors and he obtained. Very clear introduction to ZK, to PCPs, and (as an aside) to the unexpected timeline of a research project. youtu.be/bO6NT0XIAWE @tcsplus.bsky.social
- I obviously can't claim I understand the full result and its proof, but Tom's talk was so good he made me believe I *could* :)
- Bad news: Overleaf is down. The only way to edit LaTeX known to humankind!
- [Not loaded yet]
- [Not loaded yet]
- "When Pollock meets Pollard"
- Looks like I'll have to wear a buttoned-down shirt tomorrow. What did I even become an academic for?!
- [Not loaded yet]
- Thankfully, I was only going to start writing tomorrow.
- [Not loaded yet]
- I think the secretarial pool is down
- I believe, some day, we will find a way to edit text without a stable internet connection and third-party servers. One day.
- Is there work making formal connections between pseudorandomness (the computational complexity notion) and chaotic systems? Any pointers welcome. #TCSSky
- [Not loaded yet]
- That looks intriguing (and completely outside of my comfort zone). Thanks, I'll try and have a look!
- "Theoretical ML papers do not tell researchers the identity-covariance gaussians exist. Researchers already know that identity-covariance gaussians exist. Theoretical ML papers tell researchers the identity-covariance gaussians can be killed."
- [Not loaded yet]
- I was thinking more along the lines of "understood well enough, so that we can move on" :)
- (G.K. Chesterton, on spherical high-dimensional dragons)
- [Not loaded yet]
- Interestingly, the same poem captures the view of both the students taking the final, and of the faculty member having to write it.
- [Not loaded yet]
- Thanks, I'll have a look! Going directly to the quantum realm—that escalated quickly.
- ⚠️ Important for #NeurIPS2025 authors: remember that second-order optimization papers CANNOT be submitted under the CC BY-NC-ND 4.0 license. It requires "no derivatives."
- [Not loaded yet]
- Now that's just rude, David.
- [Not loaded yet]
- The dreams. They come true!
- (genuine question, not just because I'd love to write a paper called "Lorenz extractor")
- [Not loaded yet]
- [Not loaded yet]
- Also, that was a pun.
- The NeurIPS requirements state that "the main text of a submitted paper is limited to nine content pages" which clearly means that authors unhappy with their results get unlimited space
- (I should stop wasting time with these jokes and actually write that paper, shouldn't I.)
- Had forgotten about this paragraph from Borges' Library of Babel, on the importance of a shared context and vocabulary, and how we usually take it for granted.
- Famous TV series, but as programming language tutorials - Tales from the Loop: learn about control flow structures - Master of None: theory and practice of built-in data types - Special Ops: all about &, *, and . in C - True Crimes: common misuses of Boolean operators
- Succession: a friendly introduction to the Peano axioms
- [Not loaded yet]
- 📌
- The tutorials, workshops, and community events for #COLT2025 have been announced! Exciting topics, and impressive slate of speakers and events, on June 30! The workshops have calls for contributions (⏰ May 16, 19, and 25): check them out! learningtheory.org/colt2025/ind...
- [Not loaded yet]
- These are the links for the workshops (the tutorial links are not up yet). They work here – are those the ones you tried?
- Workshops at #COLT2025, CFPs and deadlines: - Theory of #AI for Scientific Computing tasc-workshop.github.io#cfp (⏰ May 16) - Foundations of Post-training fopt-workshop.github.io/cfp/ (⏰ May 19) - Predictions and Uncertainty vaidehi8913.github.io/predictions-... (⏰ May 25)
- Thank you to all of you who submitted a proposal, and a huge thanks to the workshop, tutorial, and community event organizers! Please spread the word about these, and see you at #COLT2025!
- Not my area, but if you're interested in researchers doing innovative, groundbreaking research in #HCI (on-skin wearable devices! Growing interfaces out of fungi!) then look no further than my colleague Anusha Withana at @sydneycompsci.bsky.social. He's fantastic! www.sydney.edu.au/engineering/...
- Just a shoutout after an evening talking about our respective research.
- First preprint by Kenny Chen, co-advised by Julián Mestre! We study online #algorithms with predictions, where the (unreliable) prediction is a proba distribution claimed to capture the distribution of input elements. Applications to secretary problem and metric matching! arxiv.org/abs/2505.04949
- Work led and driven by Kenny, who is amazing but does not have a website to link yet—because nobody listens to me :)
- Dear Dr. Canonne, We are sorry to inform you that your application to become pope was unsuccessful. We received a record number of submission this year, and the selection process was incredibly competitive. We encourage you to apply in future rounds. Sincerely, The POPE 2025 Committee Chairs
- [Not loaded yet]
- IV III IV I
- The pseudoscience people must be fuming. Everyone is reposting that the new pope is a maths major, but nobody mentions he's a Leo
- "We choose to go to the moon, and do the other things, not because" R1: I am not convinced by the practical relevance of the task. The authors should expand on why their focus is on the moon and not, eg, Mars. R2: Please expand on "the other things." Absent this, I am not able to assess the work.
- R3: Please cite and discuss Verne, J. (1865).
