Deep Learning

The BICI Lab does extensive work with deep learning in our research. For more details, see the Techniques we use.

Deep Learning Network

The purpose of the Deep Learning Network is to unite researchers across Imperial College London working on deep learning by facilitating the sharing of knowledge and experience, and expanding this to the wider deep learning community. There is no formal membership and all are welcome to attend meetings. You may subscribe to the mailing list (for events and general-interest postings) or Slack (for reading groups and general chat).

The group was founded by Kai in 2014 and is run with the help of volunteers both inside and outside of the university; it is not officially endorsed by Imperial College London. For more information about current events or to be added to Slack please contact Pierre.

Reading Group

The reading group happens regularly on Tuesdays at 18:00 in the Data Science Institute, South Kensington campus; we utilise room 1004 or 1009f depending on availability. All reading groups are listed in the calendar below (which can also indicate when the reading group is not running).



2017-04-12Josh Gordon (Google)What’s New in TensorFlow?Slides
2017-09-15Andrew Brock
(Heriot-Watt University/University of Edinburgh)
SMASH: One-Shot Model Architecture Search through HypernetworksSlides
2017-09-08Andrew Trask
(University of Oxford/DeepMind)
Proposing Encryption and Federation for Safe AISlides
2017-09-01Katja Hofmann
(Microsoft Research)
Minecraft as Playground and Laboratory for Artificial IntelligenceSlides
2017-08-18Yang You
(UC Berkeley)
Large-Batch DNN TrainingSlides
2017-08-08Wenzhe Shi
(Magic Pony Twitter)
Neural Networks for Image and Video Super-ResolutionSlides
2017-07-03Alex Kendall
(University of Cambridge)
Geometry in Deep Learning for Computer VisionSlides
2017-05-26Martín Arjovsky
(New York University)
On Different Distances Between Distributions and Generative Adversarial Networks
Slides (organised with the Creative AI Meetup)


These are currently being updated…