Departmental Seminar - Imprecise Probabilistic Machine Learning - Being Precise about Imprecision | Dr. Michele Caprio
| Dates: | 5 February 2025 |
|---|
| Times: | 14:00 - 15:00 |
|---|
| What is it: | Seminar |
|---|
| Organiser: | Department of Computer Science |
|---|
| How much: | Free |
|---|
| Who is it for: | University staff |
|---|
| Speaker: | Dr. Michele Caprio |
|---|
|
This talk is divided into two parts. I will first introduce the field of Imprecise Probabilistic Machine Learning, from its inception to modern-day research and open problems, including motivations and clarifying examples. In the second part, I will present Interval Deep Evidential Classifications (IDEC), a novel approach to address Uncertainty Quantification (UQ) in classification tasks. IDEC leverages an interval of evidential predictive distributions, allowing us to avoid overfitting to the training data and to systematically assess both epistemic and aleatoric uncertainties. When those surpass acceptable thresholds, IDEC has the capability to abstain from classification and flag an excess of epistemic or aleatoric uncertainty, as relevant. Conversely, within acceptable uncertainty bounds, IDEC provides a collection of labels with robust probabilistic guarantees. IDEC is trained using a loss function that draws from the theory of evidence. It overcomes the shortcomings of previous efforts and extends the current evidential deep learning literature. We also demonstrate empirically the competitive performance of IDEC in classification with abstention and out-of-distribution (OOD) evaluation settings, showcasing its effectiveness on benchmark datasets.
Speaker
Dr. Michele Caprio
Role: Lecturer (Assistant Professor) in Machine Learning
Organisation: The University of Manchester
Biography: Michele is a Lecturer (Assistant Professor) in Machine Learning at The University of Manchester. He obtained his PhD in Statistics from Duke University and worked as a postdoctoral researcher in the Department of Computer and Information Science of the University of Pennsylvania. His general interest is probabilistic machine learning, and in particular the use of imprecise probabilistic techniques to investigate the theory and methodology of uncertainty quantification in Machine Learning and Artificial Intelligence. Recently, he won the IJAR Young Researcher and the IMS New Researcher Awards, and he was elected member of the London Mathematical Society.
Travel and Contact Information
Find event
Kilburn_TH 1.3
Kilburn Building
Manchester
