52 min
Seminar, Hosted by One World Approximate Bayesian Inference (OWABI) Seminar
Post-Bayesian Machine Learning
University College London
Recording
Abstract
The machine learning community has been developing inference procedures with Bayesian characteristics that go beyond Bayes' rule as an epistemological principle. This talk explains why these efforts are needed and the forms they take, tracing the field's most important milestones through the speaker's own contributions and the challenges that lie ahead. It presents success stories of generalised Bayesian inference and the new opportunities that open up once orthodox Bayesian procedures are left behind.
Topics
- generalised Bayesian inference, broader topic: Bayesian inference
- bayesian machine learning
- inference procedures, search World Wide
- Bayesian characteristics, search World Wide
- Bayes' rule, search World Wide
- epistemological principle, search World Wide
- orthodox Bayesian procedures, search World Wide
- Bayesian procedures, search World Wide
- inference