Study of canonical and block term decompositions of higher-order tensors: uniqueness, algebraic and optimization based algorithms. KU Leuven
In many applications signals or data vary with respect to several parameters (such as spatial coordinates, velocity, time, frequency, temperature, etc.) and are therefore naturally represented by higher-order arrays of numerical values, which are called higher-order tensors. Decompositions of signals or data into simple interpretable terms correspond to the decomposition of the tensor into a sum of simple (described with small number of ...