AASS Seminar - Applied Machine Learning using Conformal Prediction - what is it good for?

The research centre AASS arranges a seminar with Ulf Norinder, MTM, Örebro University.

Remote attendance: https://oru-se.zoom.us/j/65096623293

Abstract

Conformal Prediction (CP) is a machine learning framework that provides reliable predictions originating from a statistical model given that investigated data is exchangeable. One of the advantages of using CP is that the user may set the error rate and the framework will deliver this set error rate (% errors). The talk will present the application of Conformal Prediction, introduced by Vovk, Gammerman and Shafer in 2005, and give an overview of CP. Examples from areas of research such as customer reviews, image analysis, drug design, outlier/anomaly detection and others, where reliable uncertainty estimation is desirable or crucial, will demonstrate the utility of using CP.