Research and Engineering of Robust Machine Learning Systems
Tom Diethe, Applied Science Manager, Amazon AWS.
It can now be argued that Machine Learning has become ubiquitous in many industry sectors such as retail, supply chains, advertising, and media. If we are to avoid the so-called “technical debt” associated with deploying ML systems, we need to ensure that they are efficient in long-term usage, robust to changes in the environment, and that potential errors are discoverable. Furthermore, if the data that is collected for model training and prediction is collected from individuals, systems that protect the privacy of individual data in the face of potential adversaries are paramount. In this talk I’ll cover some recent research that aims to tackle these issues, and outline some engineering-based solutions to some of these that are being made available through the AWS SageMaker cloud Machine Learning services.
About the speaker
Tom is an Applied Science Manager in Amazon AWS, based in Cambridge UK. During his time at Amazon, he has led Machine Learning teams in Supply Chain Optimization Technologies, Alexa Shopping, and is now leading a research team in AWS. Tom was formerly a Research Fellow for the “SPHERE” project at the University of Bristol, designing a platform for eHealth in a smart-home context, which was deployed into homes throughout Bristol. Tom's passions lie in probabilistic methods for machine learning, including online and continual learning, privacy aware learning, time-series modelling, unsupervised, active, and transfer learning approaches. He has a Ph.D. in Machine Learning applied to multivariate signal processing from UCL, and was employed by Microsoft Research Cambridge where he co-authored a book titled “Model-Based Machine Learning”, (see http://www.mbmlbook.com). He also has had positions at QinetiQ and the British Medical Journal. He is a fellow of the Royal Statistical Society and committee member of the Computational Statistics and Machine Learning Section, and a member of the IEEE Signal Processing Society.
Register for this event
About the seminar series
The Jean Golding Institute has teamed up with the Heilbronn Institute for Mathematical Research to showcase the latest research in Data Science - methodology with roots in Mathematics and Computer Science with important applied implications.
Our seminar series features a range of internationally regarded high-profile speakers on topics that will be relevant to a broad audience.