Result: Randomized algorithms for coupled decompositions

Coupled matrix decompositions appear as a natural extension of matrix decompositions. They find their application in different engineering disciplines where coupled decompositions are used as a tool for data fusion. We explore randomization techniques that can be used to solve this problem. In this talk we are going to discuss the adaptation of the randomized algorithms for the low-rank approximation of one matrix to the case of two (or more) matrices, with the main focus on two things, approximation error and running time of the algorithms. Different numerical examples will be presented, both on the synthetic and on the real-world data.