Efficient computation of the PARAFAC2 decomposition
Yao Cheng, Martin Haardt · 2019
The PARAFAC2 decomposition is regarded as a promising multi-linear signal processing tool in a variety of scientific fields. Designing efficient computation algorithms for PARAFAC2 is a long-standing topic. In this contribution, we propose to incorporate a dimension reduction in direct fitting-based schemes for the computation of PARAFAC2. Extensive simulations show that the proposed algorithms are able to achieve the same performance in terms of the residual as their counterpart schemes in the literature, while requiring a significantly reduced computation time.