Tensor-based algorithms for machine learning

An important research trend is the transition from vector and matrix based mathematical engineering to generalizations that make use of higher-order tensors. This trend manifests itself across many disciplines. Tensor algorithms have recently revolutionized scientific computing in high dimensions, enabling the reliable computation, in minutes or hours, of functions of which the number of unknown values “exceeds the number of atoms in the observable universe”. In machine learning and big data analysis, the potential still largely has to be unleashed. In this project we will develop numerical tensor-based algorithms for large-scale machine learning applications.