【数学学院】Data-Driven Modeling for (Semi-) Supervised Learning

摘要: Supervised and semi-supervised learning are central tasks in machine learning, where

statistics and deep learning techniques have being playing a central role. More recently,

machine learning has attached more attention of mathematicians as well. A popular trend in

mathematical modeling for (semi-) supervised learning and many other machine learning tasks

is to combine data-driven, e.g. learning, with mathematical modeling. We shall refer to this

type of modeling as “data-driven modeling”. In this talk, I will present some of our past

and recent work on data-driven modeling by combining mathematical modeling with statistical

and deep learning for both supervised and semi-supervised learning, with applications in

image restoration and recognition.