The browser you are using is not supported by this website. All versions of Internet Explorer are no longer supported, either by us or Microsoft (read more here:

Please use a modern browser to fully experience our website, such as the newest versions of Edge, Chrome, Firefox or Safari etc.

ROC Curve Explained - How smart are classifying AI:s?

Recording from AI Lund lunch seminar 9 September 2022

Topic: ROC Curve Explained - How smart are classifying AI:s?

Speaker: Kalle Åström, Professor, Mathematical Imaging Group, Lund University - Coordinator AI Lund

When: 7 September at 12.00-13.15

Where: Online

Spoken language: English


One of the many driving forces behind the ongoing AI hype was when ML-based classification systems began to perform better than humans in certain areas. For example, to recognize unique human individuals in photos. Graphically, this can be illustrated with so-called ROC curves. 

The ROC curve can be used to show how smart an AI system is when working with binary either-or problems. During the seminar, Kalle Åström tells how the ROC curve works and what is required for it to be reliable. He then discusses how to evaluate classifiers that work with more than two categories.