Predicting with confidence using conformal prediction
Speaker: Ulf Norinder, Senior Research Specialist at the Swedish Toxicology Sciences Research Center (Swetox/KI)
Title: Predicting with confidence using conformal prediction
Advantages of machine learning (ML) derived prediction models using the Conformal prediction (CP) framework has recently gained increased attention in various domains.
In this talk examples from the life science domain, e g. toxicological effects or biological activities, related to effectively detecting the minority class in highly imbalanced dataset, without employing balancing techniques such as partitioning, over- or under-sampling, will be presented in order to highlight the utility of CP derived classification models for these types of datasets.
Bio Ulf Norinder is Senior Research Specialist at the Swedish Toxicology Sciences Research Center (Swetox/KI) and Affiliated Researcher at the Dept. of Computer and System Sciences at Stockholm University. His research interests include development of in silico models for predictions of toxicological end points, quantitative risk assessment as well as pattern recognition with special emphasis on multivariate data analysis and machine learning. He is editorial board member of Perspectives in Drug Discovery and Design.