Bayesian active learning with Gaussian processes


Details
John Reid (http://johnreid.github.io) will present on Bayesian active learning with Gaussian processes.
Optimising stem cell differentiation is an important control problem in biomedicine. We use Gaussian processes to model the unknown underlying differentiation dynamics. We find that our Bayesian active learning approach (a.k.a. optimal experimental design) performs remarkably well after very few iterations and that correct quantification of uncertainty is vital in this context.
Note: For the first time we will meet at the Biscuit Factory, Block L, in Bermondsey.
Many thanks to Eyal and Richard at LabGenius (https://www.labgeni.us) and Luzana at Entrepreneur First (https://www.joinef.com) for supporting the event and providing the venue.
Agenda
18:30 - 19:00 Arrival and networking
19:00 - 20:00 Presentation, Q&A
20:00 - 20:30 Networking & discussion, moving on to a pub nearby

Bayesian active learning with Gaussian processes