Title:
Recent unsupervised and semi-supervised machine learning approaches to healthcare analytics
Talk: This talk will outline some recent machine learning approaches within healthcare analytics to tackle the singular problem represented by the key challenge to learn predictive models under the constraints of having available no or only a few labeled samples (unsupervised and semi-supervised learning). Furthermore, in cases where a few labels indeed are available, these may be noisy, and the samples/patients may belong to several diagnosis categories at the same time (multi-label learning). The new approaches are illustrated within a decision support context in different settings using patient data from electronic health records, including prediction of postoperative adverse events as well as categorization of patients potentially suffering from multiple chronic diseases.
Bio: Robert Jenssen is Professor and Head of the Machine Learning Research Group http://machine-learning.uit.no at UiT The Arctic University of Norway. He is also an adjunct professor at the Norwegian Computing Center, Oslo, Norway. Jenssen’s research interests are in the development of new machine learning algorithms, including deep learning, kernel machines, graph-based learning, and information theoretic learning, with particular focus on advancing data-driven healthcare analytics. He has had several long-term research stays abroad, at the University of Florida, at the Technical University of Berlin, and at the Technical University of Denmark. He was previously adjunct professor at the Norwegian Center for e-Health ResearchHe is an associate editor of the journal Pattern Recognition, he serves on the IEEE Technical Committee on Machine Learning for Signal Processing, is the president of the Norwegian IAPR Association, and he serves on the general board of IAPR. Jenssen is the General Chair of the annual Northern Lights Deep Learning Workshop (NLDL) http://nldl.org.
