Conformal Prediction under Ambiguous Ground Truth

Uncertainty estimation is crucial in many areas, from medical applications to self-driving cars and weather forecasting, to allow the widespread use of machine learning models.
We are excited to have David Stutz, a research scientist at Google DeepMind, in our joint Heidelberg.ai / NCT Data Science Seminar series. In this online seminar, David Stutz will talk about conformal prediction, a method to give rigorous uncertainties to machine learning models, and how to extend it to cases where even the ground truth data is uncertain.

We look forward to your participation, as this seminar will equip us with the knowledge to enhance the safety of our machine-learning models, making them even more reliable.

Abstract

In safety-critical classification tasks, conformal prediction allows to perform rigorous uncertainty quantification by providing confidence sets including the true class with a user-specified probability. This generally assumes the availability of a held-out calibration set with access to ground truth labels. Unfortunately, in many domains, such labels are difficult to obtain and usually approximated by aggregating expert opinions. In fact, this holds true for almost all datasets, including well-known ones such as CIFAR and ImageNet. Applying conformal prediction using such labels underestimates uncertainty. Indeed, when expert opinions are not resolvable, there is inherent ambiguity present in the labels. That is, we do not have ``crisp'', definitive ground truth labels and this uncertainty should be taken into account during calibration. In this paper, we develop a conformal prediction framework for such ambiguous ground truth settings which relies on an approximation of the underlying posterior distribution of labels given inputs. We demonstrate our methodology on synthetic and real datasets, including a case study of skin condition classification in dermatology.

Biography

David Stutz is a researcher and engineer with more than 7 years of AI research experience and 12 years of software engineering experience across academia and industry. Currently, he is a senior research scientist at Google DeepMind focusing on uncertainty estimation, robustness and safety evaluation of generative AI, both in vision and language applications. He helped develop and ship the first large-scale image and audio watermarking system called SynthID. In 2022, he finished his PhD at the Max Planck Institute for Informatics where he worked on adversarial robustness, quantization and uncertainty estimation with deep neural networks, focused on computer vision applications. His PhD was awarded the dissertation award of German Association for Pattern Recognition in 2023, an outstanding paper award at the CVPR 2021 CV-AML workshop, a Qualcomm Innovation Fellowship in 2019 and he was selected to participate in the Heidelberg Laureate Forum twice in 2019 and 2023 with an Abne Grant from the Carl Zeiss Foundation. Prior to his PhD, he finished his MSc and BSc at RWTH Aachen University, being supported by a Germany Scholarship and awarded the STEM Award IT and RWTH Aachen’s Springorum Denkmünze in 2018.

Event Info

Please help us plan ahead by registrating for the event via Meetup:
Event Registration .

What?	Conformal Prediction under Ambiguous Ground Truth
Who?	David Stutz, Research Scientist at Google DeepMind
When?	May 23rd 2024 @ 4pm
Where?	Zoom
Registration	meetup event-site

---