Assistant Professor & Faculty Fellow at NYU.
Probabilistic Machine Learning & RL.
Prev: Oxford + Yale. He/him.
Named Tensor Notation (TMLR version, https://arxiv.org/abs/2102.13196)
A rigorous description, opinionated style guide, and gentle polemic for named tensors in math notation.
* Macros: https://ctan.org/tex-archive/macros/latex/contrib/namedtensor
Named Tensor Notation is an attempt to define a mathematical notation with named axes. The central conceit is that deep learning is not linear algebra. And that by using linear algebra we leave many technical details ambiguous to readers.
Aalto University is looking for Assistant Professors in Computer Science. Excellent place, with opportunities to work in
@FCAI and with ELLIS. "We welcome applications in all areas of computer science" while means also #ml . https://www.aalto.fi/en/open-positions/assistant-professors-computer-science DL Jan 15, 2023
Feel free to ask for more information from all faculty, including me.
May God grant me the confidence of a large language model.
📣 You can now find *V-D4RL*, a benchmarking suite for offline RL from pixels, on #huggingface:
https://huggingface.co/datasets/conglu/vd4rl 🚀
Highlights:
💥 New D4RL-style visual datasets!
💥 Competitive baselines based on Dreamer and DrQ!
💥 A set of exciting open problems!
This is joint work with @conglu, Phil Ball, @jparkerholder, @maosbot, and @yeewhye !
Now time for a first research post...
No better time to start on offline RL from pixels! V-D4RL is now on #huggingface at https://huggingface.co/datasets/conglu/vd4rl
💥 New D4RL-style visual datasets!
💥 Competitive baselines based on Dreamer and DrQ!
💥 A set of exciting open problems!
Time for a very late introduction, I'm a 4th year PhD student at the University of Oxford interested in deep reinforcement learning, generative modelling, and Bayesian methods!
Most lately, been thinking about effective ways to automate reinforcement learning (PBT, HPO) and how to extend use cases for offline reinforcement learning (learning from pixels, generalizing to unseen environments)!
Always v. v. happy to chat :)
Seasonal reminder: Alcohol is a completely legal drug that destroys lives, relationships and families. Don't make it hard for people who are trying to stay sober to enjoy festivities and functions.
(The author of the image is https://www.madsahara.com/illustration/)
Want to do a PhD in ML? At OATML Oxford we work both on core ML methodology, as well as on applications of ML in lots of interesting domains. Deadline for fully funded studentships is Friday 9 December 2022
https://www.ox.ac.uk/admissions/graduate/courses/dphil-computer-science
[13/N] Thank you to my co-first author Tim G. J. Rudner (@timrudner), co-authors from OATML and @google and the many other collaborators who made this work possible!
[12/N] For example, in “Plex: Towards Reliability using Pretrained Large Model Extensions” @dustinvtran et al.), we evaluate the performance of pretrained models on RETINA. (https://arxiv.org/abs/2207.07411)
[10/N] To enable future research on reliability in safety-critical settings, the RETINA Benchmark is open-sourced as part of Uncertainty Baselines:
https://github.com/google/uncertainty-baselines
Announcing the public release of the #̶N̶e̶u̶r̶I̶P̶S̶2̶0̶2̶2̶ #NeurIPS2021 (😅) RETINA Benchmark:
A suite of tasks evaluating the reliability of uncertainty quantification methods like Deep Ensembles, MC Dropout, Parameter- and Function-Space VI, and more.
Paper: https://arxiv.org/abs/2211.12717
Code+Checkpoints: https://rebrand.ly/retina-benchmark
#NewPaper #arxiv #PaperThread
🧵 below👇🏾 [0/N]
@lawmurray Would you use it also for discrete y? It's fine to introduce new notation, but I also need to teach my students to understand existing common conventions. The attached image shows the existing common way to be more specific about p(y|theta), but these get complicated when y or theta is a combination of discrete and continuous variables
@avehtari @lawmurray it gets even more confusing when we do not consider parameters to be random variables (e.g., in the frequentist case), but people still "condition over them" in notation.
And when they are random variables, additional care shall be put when we are conditioning over a zero-probability event, e.g., when parameters are continuous. In that case for discrete observations we are describing a mixed continuous-discrete distribution!
PhD students: Want to work with Mary Gray, Nancy Baym, Tarleton Gillespie, and me at Microsoft Research on sociotechnical projects? We are hiring both interns and postdocs. For more info, see: https://socialmediacollective.org/
(Please boost/share with potential PhD candidates)
Very excited to announce that I am looking for #PhD students to work with @ umich CSE. If you're interested in #causality + #machinelearning and would like to spend your PhD years NOT losing half your income on rent, I strongly encourage you to apply to our PhD program
📺 I started a new video series on primitive rules for #automaticdifferentiation :https://www.youtube.com/watch?v=PwSaD50jTv8&list=PLISXH-iEM4Jn3SEi07q8MJmDD6BaMWlJE
starting at scalar rules, continuing with vector/array rules, and finally some results from using the implicit function theorem.
Primitive rules build the basis for automatically differentiating through arbitrary computer programs.
New video to be released every three days :)
@chelseaparlett Various US gov't agencies are using auto ML for safety critical applications (e.g., in defense, intelligence, treasury), and I'm researching potential safety risks of autoML in those domains with the Center for Security & Emerging Technology at Georgetown. While education on ML/autoML would be unambiguously useful, better regulation of autoML may also be an avenue to avoid harm. What minimum level of understanding of ML should do you think users of autoML should have have?
@mundt_martin super interesting workshop!
RT @MisterRatt
Twitter Blue is going about as well as everyone predicted, and it's an amazing spectacle to watch. Like a train crash filled with glitter.
