Making sense of the world one model at a time ~ PhD in Machine Learning and Statistics from @UniofOxford
Oxford, England
Joined May 2018
- Tweets525
- Following275
- Followers577
- Likes4.7K
Pinned Tweet
Do you keep hearing about this awesome algorithm called Hamiltonian Monte Carlo (HMC) and want to know more?
Look no further! Here is my beginners guide to HMC with code and pretty pictures! If you like it, please RT :)
#bayesian #MachineLearning #MCMC
bayesianbrad.github.io/posts…
Bradley Gram-Hansen 🌌 retweeted
🧵1/7 Interested in who will become the next US President and how recent news events influenced the chances of each candidate? Let's dive into how election odds can offer insights. 👀
Bradley Gram-Hansen 🌌 retweeted
Thank you @icmlconf! Incredibly humbling experience to reflect upon all the amazing work by the ML community that has followed from our little paper. I am lucky to have @wellingmax as a friend and collaborator over the years. It was a blast working on the paper :)
Bradley Gram-Hansen 🌌 retweeted
My team at Microsoft Research Cambridge is hiring a #MachineLearning intern! If you're working in the exciting areas of meta-learning/uncertainty quantification and are passionate about real-world problems, then apply by 15 Aug - bit.ly/3hvzVH4 🤖@MSFTResearchCam
Bradley Gram-Hansen 🌌 retweeted
We're "go" for launch to send our machine learning model to space on a @SpaceX #Falcon9 today in a few hours!!! This is the hard work of the team @gonzmg88 @drjoshvm Lewis Smith @_silviu_oprea_ @guy_schumann @yaringal me and @DJBackes You can watch spacex.com/launches/
Falcon 9 and 88 spacecraft are vertical on pad 40 ahead of today’s launch. The 58-minute window opens at 2:56 p.m. EDT, and the webcast will go live about 15 minutes before liftoff → spacex.com/launches
Bradley Gram-Hansen 🌌 retweeted
4 days until the Robust Video Scene Understanding Workshop at CVPR!
eval.vision.rwth-aachen.de/r…
8 exciting invited speakers including @FidlerSanja @kkitani @judyfhoffman @WeidiXie, Katerina Fragkiadaki, Philipp Krähenbühl, Michael Felsberg and Lorenzo Torresani.
+ MUCH MORE..
🧵👇
Bradley Gram-Hansen 🌌 retweeted
super excited that our latest work examing the effect of (self-reported) mask **wearing** in community settings on COVID-19 transmission is now out ! check out the thread below for a great summary
How well do masks work against COVID transmission, at the population level?
In our new preprint, we find that an entire population wearing masks is linked to a reduction in R of 24.6% [6%, 43%].
medrxiv.org/content/10.1101/…
Bradley Gram-Hansen 🌌 retweeted
New paper arxiv.org/abs/2106.07780 on Domain Adaptation, with Toan Tran, @yaringal, Phil Torr, and @atilimgunes [1/n]
Bradley Gram-Hansen 🌌 retweeted
I have incredible news that something I worked on is going to SPACE (!) on the upcoming @SpaceX #Falcon9 launch Transporter-2 from SLC-40 Cape Canaveral 🚀🌌😀 More details to follow! @FDL_AI @D_Orbit Our paper: nature.com/articles/s41598-0…
Bradley Gram-Hansen 🌌 retweeted
After a long time, the 8th blog post is out! This time I discuss neural compression with deep generative modeling.
Post 🤓: jmtomczak.github.io/blog/8/8…
Code 💻: github.com/jmtomczak/intro_d…
Bradley Gram-Hansen 🌌 retweeted
Excited to announce that we have released the code for Deep Deterministic Uncertainty (DDU) including a standardized way to generate and use Dirty-MNIST!
💻Code github.com/omegafragger/DDU
0⃣-9⃣Dirty-MNIST: blackhc.github.io/ddu_dirty_…
📄Updated Paper: arxiv.org/abs/2102.11582
Did you know that you can beat Deep Ensemble Uncertainty with a single deterministic net?
We prove that softmax nets can't normally capture epistemic uncertainty, but with an appropriate inductive bias any pre-trained net can implicitly capture uncertainty
arxiv.org/abs/2102.11582
Bradley Gram-Hansen 🌌 retweeted
I think conference registration can easily add an optional “would you like to sponsor students And how many?” Question. And then offer free registrations.
BTW many ML conferences have additional budget to sponsor students (eg at ICLR), but this needs to be expanded.
This Post is from an account that no longer exists. Learn more
Bradley Gram-Hansen 🌌 retweeted
Happy to share a repo that contains 36 publicly available datasets for African languages across 8 NLP tasks that I have been putting together for several months. Please, feel free to DM me or do a pull request if there is any related dataset I missed out.
github.com/Andrews2017/afric…
ALT Photo from Wikipedia.
Bradley Gram-Hansen 🌌 retweeted
Interested in Kernel Methods, Causal Inference and Uncertainty quantification?
In our newest work we introduce BayesIMP: Uncertainty Quantification for Causal Data Fusion !
arxiv.org/abs/2106.03477
Big thanks to @Chau9991 @javiergonzh @yeewhye @sejDino 1/n
Bradley Gram-Hansen 🌌 retweeted
🗞New Paper🗞
🤖🧪Self-Attention Between Datapoints: Going Beyond Individual Input-Output Pairs in Deep Learning 🧪🤖
Huge thanks to @neilbband* as well as @clarelyle, @AidanNGomez, @tom_rainforth, @yaringal, and @OATML_Oxford !
Introducing 🚀Non-Parametric Transformers🚀 1/
ALT Overview of the Non-Parametric Transformer. (a) The input dataset and mask matrix are stacked and (b) linearly embedded for all datapoints independently. NPT then applies (c) Attention Between Datapoints across all n samples of hidden dimension h = d · e. (d) Attention Between Attributes then attends between the attributes for each datapoint independently. We repeat steps (c) and (d) and obtain a final prediction from a separate linear projection.
Bradley Gram-Hansen 🌌 retweeted
A related gem from Grzeszczuk, Terzopoulos and Hinton. NeuroAnimator: Fast Neural Network Emulation and Control of Physics-Based Models, SIGGRAPH 1998
cs.toronto.edu/~hinton/absps…
Gargan & Neelamkavil, Approximating Reflectance Functions using Neural Networks, EGSR 1998
books.google.ca/books?id=OJS…
That's about 23 years ahead of its time. Just imagine what you could learn at online EGSR'21 this year...
Bradley Gram-Hansen 🌌 retweeted
Gargan & Neelamkavil, Approximating Reflectance Functions using Neural Networks, EGSR 1998
books.google.ca/books?id=OJS…
That's about 23 years ahead of its time. Just imagine what you could learn at online EGSR'21 this year...
Bradley Gram-Hansen 🌌 retweeted
Had some more printed, so still have a few copies! DM your address if you want one, I only charge for shipping and even that is free if you can’t afford it.
This quoted post is unavailable.
Bradley Gram-Hansen 🌌 retweeted
Contrastive learning aims to learn representation such that similar samples stay close, while dissimilar ones are far apart. It can be applied to supervised / unsupervised data and has been shown to achieve good results on various tasks.
📚 A long read: lilianweng.github.io/lil-log…
Bradley Gram-Hansen 🌌 retweeted
Now published in JMLR (jmlr.org/papers/v22/20-1369.…): extended and improved risk analysis of random Fourier features building on our ICML 2019 paper which received a Best Paper Honorable Mention.
With @lzy_michael @jeanfrancois287 @DinoOglic
"Towards a Unified Analysis of Random Fourier Features", by Zhu Li, Jean-Francois Ton, Dino Oglic, Dino Sejdinovic. jmlr.org/papers/v22/20-1369.…