High Throughput Bayesian Optimisation

Add to your list(s) Download to your calendar using vCal

• Victor Picheny (None / Other)
• Thursday 20 April 2023, 11:00-12:00
• Seminar Room 2, Newton Institute.

If you have a question about this talk, please contact nobody.

DDE - The mathematical and statistical foundation of future data-driven engineering

Bayesian optimisation excels in small data regimes, but its large computational overhead and the cubic cost of vanilla Gaussian Process models makes it impractical as soon as the data size reaches thousands of points. In this talk, we will expose some of the work developed at Secondmind, fueled by real-world applications from the automotive industry, to expand the capability of Bayesian optimisation to tackle high data regimes (typically thousands to millions of observations). Leveraging sparse variational GP models and Thompson sampling strategies, we will demonstrate that Bayesian optimisation can enjoy both strong theoretical guarantees and empirical performance in this context.

This talk is part of the Isaac Newton Institute Seminar Series series.

This talk is included in these lists:

• All CMS events
• Featured lists
• INI info aggregator
• Isaac Newton Institute Seminar Series
• School of Physical Sciences
• Seminar Room 2, Newton Institute
• bld31

Note that ex-directory lists are not shown.