PRECISE Seminar Talk “Evaluating and Calibrating AI Models with Uncertain Ground Truth”

I had the pleasure to present our work on evaluating and calibrating with uncertain ground truth at the seminar series of the PRECISE center at the University of Pennsylvania. Besides talking about our recent papers on evaluating AI models in health with uncertain ground truth and conformal prediction with uncertain ground truth, I also got to learn more about the research at PRECISE through post-doc and student presentations. In this article, I want to share the corresponding slides.


For safety, AI systems in health undergo thorough evaluations before deployment, validating their predictions against a ground truth that is assumed certain. However, this is actually not the case and the ground truth may be uncertain. Unfortunately, this is largely ignored in standard evaluation of AI models but can have severe consequences such as overestimating the future performance. To avoid this, we measure the effects of ground truth uncertainty, which we assume decomposes into two main components: annotation uncertainty which stems from the lack of reliable annotations, and inherent uncertainty due to limited observational information. This ground truth uncertainty is ignored when estimating the ground truth by deterministically aggregating annotations, e.g., by majority voting or averaging. In contrast, we propose a framework where aggregation is done using a statistical model. Specifically, we frame aggregation of annotations as posterior inference of so-called plausibilities, representing distributions over classes in a classification setting, subject to a hyper-parameter encoding annotator reliability. Based on this model, we propose a metric for measuring annotation uncertainty and provide uncertainty-adjusted metrics for performance evaluation. We present a case study applying our framework to skin condition classification from images where annotations are provided in the form of differential diagnoses. The deterministic adjudication process called inverse rank normalization (IRN) from previous work ignores ground truth uncertainty in evaluation. Instead, we present two alternative statistical models: a probabilistic version of IRN and a Plackett-Luce-based model. We find that a large portion of the dataset exhibits significant ground truth uncertainty and standard IRN-based evaluation severely over-estimates performance without providing uncertainty estimates.

Slides Papers covered:
[1] David Stutz and Ali Taylan Cemgil and Abhijit Guha Roy and Tatiana Matejovicova and Melih Barsbey and Patricia Strachan and Mike Schaekermann and Jana von Freyberg and Rajeev Vijay Rikhye and Beverly Freeman and J. Perez Matos and Umesh Telang and Dale R. Webster and Yuan Liu and Greg S Corrado and Yossi Matias and Pushmeet Kohli and Yun Liu and Arnaud Doucet and Alan Karthikesalingam.
Evaluating AI systems under uncertain ground truth: a case study in dermatology. ArXiv, 2023.
[2] David Stutz and Abhijit Guha Roy and Tatiana Matejovicova and Patricia Strachan and Ali Taylan Cemgil and Arnaud Doucet.
Conformal prediction under ambiguous ground truth. TMLR, 2023.
What is your opinion on this article? Let me know your thoughts on Twitter @davidstutz92 or LinkedIn in/davidstutz92.