I'll be recruiting a PhD student through the NYU Data Science program (to start in Fall 2021). Would love to chat about it "at" #acl2020nlp. Email me if interested in a computational psycholinguistics ∩ deep learning PhD (even if you're not attending the conference)!

The #ICML2020 tutorial on Bayesian Deep Learning and Probabilistic Model Construction is tomorrow! July 13, 11 am - 2 pm and 9 pm - 12 am EDT icml.cc/virtual/2020/t… See you there!

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…

New work in visual imitation learning! We use commodity reacher-grabbers to both collect demonstrations in the wild, and as an end-effector for robotic manipulation. This means no domain shift b/n the demonstrations and the robot's observations! arxiv.org/abs/2008.04899 1/5

Very excited to share our newest paper combining machine learning & physics. We develop normalizing flows that impose the elaborate of symmetry groups you find in fundamental particle physics. It's a beautiful mix of math, ML, and physics arxiv.org/abs/2008.05456

I'm so excited to be presenting on "Bayesian Deep Learning" and "How do we build models that learn and generalize?" at the SMILES - Summer School of Machine Learning at SK Summer School. 10 am ET tomorrow+Mon, livestreamed on YouTube! Functional kernel learning will make an appearance. youtube.com/watch?v=i5PEMt…

Stochastic Weight Averaging (SWA) is a simple procedure that improves generalization in deep learning over Stochastic Gradient Descent (SGD). PyTorch 1.6 now includes SWA natively. Learn more from Pavel Izmailov, Andrew Gordon Wilson and Vincent: pytorch.org/blog/pytorch-1…

new blog post on evaluating the quality of representation!

A new NLP toolkit from CILVR is now online with an accompanying blog post: wp.nyu.edu/cilvr/2020/10/…. check it out!

Translation equivariance has imbued CNNs with powerful generalization abilities. Our #NeurIPS2020 paper shows how to *learn* symmetries -- rotations, translations, scalings, shears -- from training data alone! arxiv.org/abs/2010.11882 w/ Greg, Pavel Izmailov, Marc Finzi. 1/9

I had a wonderful time chatting with @krrish94 and Liam Paull at the REAL - Robotics and Embodied AI Lab Robot Learning seminar yesterday! Talk is on recorded here: youtu.be/rPrqCPJc6wU

for the 3rd year in a row, we are carbon-neutral. thanks Sam Bowman for leading the effort!

One of the coolest ideas I spent time on this last year considers a weird way to specify confounders. In most traditional causal estimation, the confounders and interventions of interest are different columns in a dataset of pre-outcome variables.

Courant Institute and Radiology at NYU are hiring postdocs in the area of learning unified representations of patients using healthcare data from multiple modalities (imaging, EHRs, sensors). An exciting opportunity to make real-world impact. Details at: apply.interfolio.com/85305

our own Kyle Cranmer is a keynote speaker at AISTATS Conference !

Fantastic lineup! Submission deadline: September 18 #ML4PS2021

Talks at TTIC presents Naomi Saphra (Naomi Saphra) of NYU (CILVR) with a talk titled "Interpreting Training." Please join us on Monday, February 27 at 11:30 am CT in Room 530 or via Panopto: buff.ly/3xINPhu