Andrew Gordon Wilson (@andrewgwils) / X

Andrew Gordon Wilson

3,308 posts

Andrew Gordon Wilson

@andrewgwils

Machine Learning Professor

New York University

cims.nyu.edu/~andrewgw

Joined September 2014

Pinned
Andrew Gordon Wilson
@andrewgwils
Mar 5, 2025
My new paper "Deep Learning is Not So Mysterious or Different": arxiv.org/abs/2503.02113. Generalization behaviours in deep learning can be intuitively understood through a notion of soft inductive biases, and formally characterized with countable hypothesis bounds! 1/12
336K
Andrew Gordon Wilson
@andrewgwils
Oct 1, 2025
I miss the days of being a PhD student, or postdoc. I would give almost anything to have multiple full days at a time, just to concentrate deeply and single-mindedly on open-ended research.
297K
Andrew Gordon Wilson
@andrewgwils
Jan 16, 2023
Does anyone else fantasize about freezing time for a few months, for thinking and learning, without expectations, responsibilities, or distractions?
254K
Andrew Gordon Wilson
@andrewgwils
Jun 29, 2025
You don't _need_ a PhD (or any qualification) to do almost anything. A PhD is a rare opportunity to grow as an independent thinker in an academic environment, rather than immediately becoming a gear in a corporate agenda. It's definitely not for everyone!
Noam Brown
@polynoamial
Jun 28, 2025
You don’t need a PhD to be a great AI researcher. Even @OpenAI’s Chief Research Officer doesn’t have a PhD.
198K
Andrew Gordon Wilson
@andrewgwils
Dec 27, 2019
Bayesian methods are *especially* compelling for deep neural networks. The key distinguishing property of a Bayesian approach is marginalization instead of optimization, not the prior, or Bayes rule. This difference will be greatest for underspecified models like DNNs. 1/18
Andrew Gordon Wilson
@andrewgwils
Oct 23, 2024
I couldn't agree more. Build your foundations: mathematics, physical intuitions, writing... these will never go obsolete. The most useful course I took in college superficially seemed the most useless: real analysis. But it changed my brain, and the way I filter the world.
Yann LeCun
@ylecun
Oct 23, 2024
Advice to students: study fundamental topics that have a long shelf life.
103K
Andrew Gordon Wilson
@andrewgwils
Oct 3, 2018
I'm extremely excited to officially announce our new library, GPyTorch, which has just gone beta! Scalable Gaussian processes in PyTorch, with strong GPU acceleration. repo: github.com/cornellius-gp/… website: gpytorch.ai
Andrew Gordon Wilson
@andrewgwils
Feb 21, 2020
Our new paper "Bayesian Deep Learning and a Probabilistic Perspective of Generalization": arxiv.org/abs/2002.08791. Includes (1) benefits of BMA; (2) BMA <-> Deep Ensembles; (3) new methods; (4) BNN priors; (5) generalization in DL; (6) tempering in BDL. With @Pavel_Izmailov. 1/19
Andrew Gordon Wilson
@andrewgwils
Jul 20, 2025
Something I noticed at ICML: for the most part, VCs are strikingly disinterested in the technical content of the conference, and the discipline at large. You could be describing what obsoletes the transformer, and they couldn’t care less.
123K
Andrew Gordon Wilson
@andrewgwils
Dec 29, 2020
There is a lot of often overlooked evidence that standard p(w) = N(0, a*I) priors combined with a NN f(x,w) induce a distribution over functions p(f(x)) with useful properties!... 1/15
Andrew Gordon Wilson
@andrewgwils
Oct 16, 2019
We're extremely excited to release our BoTorch paper! Scalable, flexible, and modular Bayesian optimization integrated with GPyTorch and @PyTorch. It's been a pleasure working with @eytan, Max, and team. paper: arxiv.org/abs/1910.06403 code: github.com/pytorch/botorch
Andrew Gordon Wilson
@andrewgwils
Jun 25, 2024
Physics is cooler than AI. Gravitational time dilation? The twin paradox? Quantum entanglement? A chatbot just doesn't compare...
218K
Andrew Gordon Wilson
@andrewgwils
Jul 27, 2020
I had a wonderful experience presenting the ICML tutorial on Bayesian deep learning and probabilistic model construction. Thank you all for your encouraging comments. The video is now online! youtube.com/watch?v=E1qhGw…
Andrew Gordon Wilson
@andrewgwils
Jun 24, 2025
Machine learning is perhaps the only field that has become less mature over time. A reverse metamorphosis, from butterfly to caterpillar.
62K