Publications

We do research in both theoretical machine learning, as well as applications of machine learning in various domains, including medical, astronomy, autonomous driving, and more.

Recent publications:

• 05/2019: Andreas Kirsch, Joost van Amersfoort, Yarin Gal , "BatchBALD: Efficient and Diverse Batch Acquisition for Deep Bayesian Active Learning".
• 05/2019: Mike Walmsley, Lewis Smith, Chris Lintott, Yarin Gal, Steven Bamford, Hugh Dickinson, Lucy Fortson, Sandor Kruk, Karen Masters, Claudia Scarlata, Brooke Simmons, Rebecca Smethurst, Darryl Wright , "Galaxy Zoo: Probabilistic Morphology through Bayesian CNNs and Active Learning".
• 05/2019: Adam D. Cobb, Michael D. Himes, Frank Soboczenski, Simone Zorzan, Molly D. O'Beirne, Atılım Güneş Baydin, Yarin Gal, Shawn D. Domagal-Goldman, Giada N. Arney, Daniel Angerhausen , "An Ensemble of Bayesian Neural Networks for Exoplanetary Atmospheric Retrieval".

Reproducibility and Code

One of our lab's missions is to contribute to the reproducibility effort. Here we provide code for our own research, as well as reproductions of works by others (e.g. ones we implemented as baselines as part of our own research).

Recent code release:

• 07/2019: Andreas Kirsch , Putting TensorFlow back in PyTorch, back in TensorFlow (differentiable TensorFlow PyTorch adapters).
• 06/2019: Andreas Kirsch, Joost van Amersfoort, Yarin Gal , Code for BatchBALD blog post and paper (active learning).
• 06/2019: Joost van Amersfoort , Reproducing the results from "Do Deep Generative Models Know What They Don't Know?".

Group Blog

Recent group blog posts:

• 02 Jul 2019: Poor generalization can be dangerous in RL!
• 24 Jun 2019: Human in the Loop: Deep Learning without Wasteful Labelling
• 14 Jun 2019: Bayesian Deep Learning Benchmarks
• 05 Jun 2019: Targeted Dropout
• 09 Apr 2019: An imPACtful, BAYESic result

Group Members

We are researchers coming from varied backgrounds, including Computer Science, Maths & Stats, Engineering, and Physics.

We come from academia (Oxford, Cambridge, MILA, McGill, U of Amsterdam, U of Toronto, Yale, and others) and industry (Google, DeepMind, Twitter, Qualcomm, and startups).

We include 3 Rhodes Scholars, 2 Clarendon Scholars, one DeepMind Scholar, with our students funded by many additional sources (AIMS CDT, Cyber CDT, and more).