Postdoctoral Researcher, Google Brain Berlin
- Date & time
- Monday, January 11, 2021 · 1:00 PM
- Location
- Online (live stream)
Abstract
Deep learning models are bad at detecting their failure, tending to make over-confident mistakes especially under distribution shift. We discuss two approaches to reliable deep learning. First, we focus on Bayesian neural networks and cast doubt on the current understanding of Bayes posteriors in deep networks, showing that they can be improved significantly through a cold posterior that sharply deviates from the Bayesian paradigm, and discuss hypotheses that could explain it. Second, we discuss ensembles: we show that the diversity of predictions can be improved by considering models with different hyperparameters, and present an efficient method that leverages hyperparameter diversity within a single model.
About the speaker
Florian Wenzel is a machine learning researcher whose work has focused on probabilistic deep learning, uncertainty estimation, and scalable inference. From 2019 to 2020 he was a postdoctoral researcher at Google Brain. He received his PhD from Humboldt University in Berlin, working with Marius Kloft, Stephan Mandt, and Manfred Opper.