报告主题：Quantification of Prediction Uncertainty using Random Fuzzy Sets

报告人：Thierry Denoeux

报告时间： 2026年7月16日 下午3点

报告地点：计算机楼A411

报告摘要：

Uncertainty quantification is essential in prediction, yet conventional probabilistic models do not always distinguish intrinsic variability from uncertainty caused by limited data, imperfect knowledge or imprecise information. This talk introduces random fuzzy sets as a unified framework for representing these different forms of uncertainty. Random fuzzy sets extend both random sets and fuzzy sets and encompass probability distributions, possibility distributions and belief functions as particular cases. After introducing their basic interpretation through belief, plausibility and contour functions, we present Gaussian random fuzzy numbers and vectors, a tractable parametric family in which randomness describes aleatory variability and fuzziness represents epistemic uncertainty. We then discuss two approaches to constructing predictive random fuzzy sets in machine learning. The first is a prototype-based approach, illustrated by the ENNreg model, in which distances to learned prototypes generate local pieces of evidence that are combined to quantify regression uncertainty. The second is a likelihood-based approach, in which relative likelihood is interpreted as a possibility distribution over model parameters and propagated jointly with random noise through the prediction equation. These approaches provide point predictions together with informative uncertainty outputs, including belief intervals, lower and upper distribution functions, and indicators of extrapolation or insufficient evidence. 

报告人简介：

Thierry Denoeux graduated from ?cole nationale des ponts et chaussées in Paris, France and earned a PhD from the same institution. He is currently a Full Professor (Exceptional Class) with the Department of Information Processing Engineering at Université de Technologie de Compiègne, France. He is the president of the Belief Functions and Applications Society. In 2019, he was appointed as a senior member of Institut Universitaire de France, and he was reconducted in 2024. His research interests concern reasoning and decision-making under uncertainty and, more generally, the management of uncertainty in intelligent systems. His main contributions are in the theory of belief functions with applications to statistical inference, machine learning and information fusion. He is the author of more than 350 papers in journals and conference proceedings and he has supervised more than 30 PhD theses. He is the Editor-in-Chief of the International Journal of Approximate Reasoning (Elsevier), and an Associate Editor of several journals including Fuzzy Sets and Systems and IEEE Transactions on Fuzzy Systems.