L
Lewis Smith
Researcher at University of Oxford
Publications - 31
Citations - 1201
Lewis Smith is an academic researcher from University of Oxford. The author has contributed to research in topics: Artificial neural network & Uncertainty quantification. The author has an hindex of 14, co-authored 31 publications receiving 817 citations. Previous affiliations of Lewis Smith include Genentech.
Papers
More filters
Posted Content
Understanding Measures of Uncertainty for Adversarial Example Detection
Lewis Smith,Yarin Gal +1 more
TL;DR: In this article, failure modes for MC dropout, a widely used approach for estimating uncertainty in deep models, are highlighted, and a proposal to improve the quality of uncertainty estimates using probabilistic model ensembles is made.
Journal ArticleDOI
The Fluorination of C−H Bonds: Developments and Perspectives
TL;DR: Overall, the field of late stage nucleophilic C-H fluorination has progressed much more slowly, a state of play explaining why C- H 18F-fluorination is still in its infancy.
Journal ArticleDOI
Galaxy Zoo: probabilistic morphology through Bayesian CNNs and active learning
Mike Walmsley,Lewis Smith,Chris Lintott,Yarin Gal,Steven P. Bamford,Hugh Dickinson,Lucy Fortson,Sandor Kruk,Karen L. Masters,Karen L. Masters,Claudia Scarlata,Brooke Simmons,Brooke Simmons,Rebecca Smethurst,Darryl Wright +14 more
TL;DR: By combining human and machine intelligence, Galaxy Zoo will be able to classify surveys of any conceivable scale on a timescale of weeks, providing massive and detailed morphology catalogues to support research into galaxy evolution.
Proceedings Article
Uncertainty Estimation Using a Single Deep Deterministic Neural Network
TL;DR: DUQ as discussed by the authors is a deterministic deep model that can find and reject out-of-distribution data points at test time with a single forward pass, based on the idea of RBF networks.
Posted Content
A Systematic Comparison of Bayesian Deep Learning Robustness in Diabetic Retinopathy Tasks.
Angelos Filos,Sebastian Farquhar,Aidan N. Gomez,Tim G. J. Rudner,Zachary Kenton,Lewis Smith,Milad Alizadeh,Arnoud de Kroon,Yarin Gal +8 more
TL;DR: A new BDL benchmark with a diverse set of tasks, inspired by a real-world medical imaging application on diabetic retinopathy diagnosis, and a systematic comparison of well-tuned BDL techniques on the various tasks concludes that some current techniques which solve benchmarks such as UCI `overfit' their uncertainty to the dataset underperform on this benchmark.