Surya Ganguli
@SuryaGanguli
Associate Prof of Applied Physics @Stanford, and departments of Computer Science, Electrical Engineering and Neurobiology. Venture Partner @a16z
One way to think about it: I like exercising - lifting some weights & running. But a crane lifts more than me, and a car goes faster than me. This takes nothing from the sheer human joy of exercise. Also fast cars add to our joy of superhuman speed. Same w/ math. And chess & go.
the openai IMO news hit me pretty heavy this weekend i'm still in the acute phase of the impact, i think i consider myself a professional mathematician (a characterization some actual professional mathematicians might take issue with, but my party my rules) and i don't think i…
Nice - we could have prompted our way to winning IMO gold on an already available model!
🚨 Olympiad math + AI: We ran Google’s Gemini 2.5 Pro on the fresh IMO 2025 problems. With careful prompting and pipeline design, it solved 5 out of 6 — remarkable for tasks demanding deep insight and creativity. The model could win gold! 🥇 #AI #Math #LLMs #IMO2025
“The U.S. answered Sputnik with bold competition not an aggrieved whine that America had been ripped off and abused.” (Tripled NSF funding in 1 year; 10x in 10; attracted talent through immigration).
What happened after Sputnik? @ATabarrok provides a striking summary: NSF funding tripled in a year and increased by a factor of more than ten by the end of the decade. The National Defense Education Act overhauled universities and created new student loan programs for foreign…
I'm at ICML presenting this work! Come by on Tuesday to hear about/chat about combinatorial generalization and creativity in diffusion models. x.com/MasonKamb/stat…
Excited to finally share this work w/ @SuryaGanguli. Tl;dr: we find the first closed-form analytical theory that replicates the outputs of the very simplest diffusion models, with median pixel wise r^2 values of 90%+. arxiv.org/abs/2412.20292
This should be very interesting and instructive. Happy to be on the advisory board of this conference for AI written and reviewed papers. Consider submitting!!
📢New conference where AI is the primary author and reviewer! agents4science.stanford.edu Current venues don't allow AI-written papers, so it's hard to assess the +/- of such works🤔 #Agents4Science solicits papers where AI is the main author w/ human advisors. 💡Initial reviews by…
📢New conference where AI is the primary author and reviewer! agents4science.stanford.edu Current venues don't allow AI-written papers, so it's hard to assess the +/- of such works🤔 #Agents4Science solicits papers where AI is the main author w/ human advisors. 💡Initial reviews by…
📢 Call for Papers: NeurReps 2025 ‼️‼️‼️ 🧠 Submit your research on symmetry, geometry, and topology in artificial and biological neural networks. Two tracks: Proceedings (9 pages) and Extended Abstract (4 pages). Deadline: Aug 22, 2025. neurreps.org/call-for-papers
btw, 1) we proved conditions for a simple version of the Platonic Representation Hypothesis for two layer linear networks back in 2019 in our @PNASNews paper: A mathematical theory of semantic development in deep nets (Fig. 11): pnas.org/doi/10.1073/pn… 2) we also showed evidence…
Fun Facts: NIH budget for research on diseases causing > 90% of human deaths is reduced by 40% Homicide deaths (whether by illegal alien or not) < 0.01% but ICE funding is increased by 300% Total ICE budget > 6 times yearly NIH budget Homicide deaths < 0.01 disease deaths
Summary Table of the massive defunding of biomedical research in the United States nytimes.com/interactive/20…
My lab’s 5-year NIH R01 grant, awarded to study gene therapy for hearing loss, was abruptly terminated. I want to share how this action has been incredibly harmful and disruptive, not just to my lab, but to the scientific process itself. 1/15
In a recent paper, the physicists Mason Kamb (left) and Surya Ganguli found that the creativity in diffusion models is a consequence of their architecture. quantamagazine.org/researchers-un…
This week in academia, a not so short🧵... 1. Staff reductions and other cost cutting measures coming to Brown University highereddive.com/news/brown-uni…
This syllabus contains ancient antediluvian wisdom from before the flood of deep learning scaling. Young people would do well to master it. E.g Graphical models are essential in AI for science where interpretability and well calibrated uncertainty quantification is needed. They…
This is the syllabus of the course @geoffreyhinton and I taught in 1998 at the Gatsby Unit (just after it was founded). Notice anything?
Honored to have had my recent work with @SuryaGanguli on the mechanisms behind creativity in diffusion models featured in this lovely article by @_webbwright for Quanta magazine!
In a recent paper, physicists used two predictable factors to reproduce the “creativity” seen from image-generating AI. @_webbwright reports: quantamagazine.org/researchers-un…
A great @QuantaMagazine article on our theory of creativity in convolutional diffusion models lead by @MasonKamb. See also our paper with new results in version 2: arxiv.org/abs/2412.20292 to be presented as an oral at @icmlconf #icml25 thx @_webbwright !
In a recent paper, physicists used two predictable factors to reproduce the “creativity” seen from image-generating AI. @_webbwright reports: quantamagazine.org/researchers-un…
No. Some (biased) counterexamples: A 2D dynamical system predicting mouse brain activity: nature.com/articles/s4158… Random projections and scaling in the brain: sciencedirect.com/science/articl… Critical dynamics triggering perception: science.org/doi/10.1126/sc… Random matrix theory and…
Theoretical neuroscience has become a series of self-referential mathematical exercises, untroubled by the messy realities of the brain. It’s now more a subfield of theoretical physics than biology, where the universe is replaced with a collection of neurons and data is optional
I think it’s even more beautiful if written as e^(i pi) + 1 = 0. Then the additive identity 0 and the multiplicative identity 1 both get brought in to play a role, in addition to differential equations (e), geometry (pi) and algebraic closure (i).
What you see below is one of the most beautiful formulas in mathematics. A single equation, establishing a relation between 𝑒, π, the imaginary number, and 1. It is mind-blowing. This is what's behind the sorcery: