0460710 20240103212406.820160714235959.9 84978100769 000379694800005 10.1109/LSP.2016.2577383 5 s. P cav_un_epca*02532121070-9908237 (2016)993-997Institute of Electrical and Electronics Engineers canonical polyadic decomposition PARAFAC tensor decomposition cav_un_auth*0101212 Stochastická informatika Department of Stochastic Informatics SI SI Department of Stochastic Informatics Tichavský Petr UTIA-B Ústav teorie informace a automatizace AV ČR, v. v. i. cav_un_auth*0274170 Phan A. H. JP cav_un_auth*0274171 Cichocki A. JP http://library.utia.cas.cz/separaty/2016/SI/tichavsky-0460710.pdf cav_un_auth*0303443 GA14-13713S GA ČR CZ Canonical polyadic decomposition (CPD), also known as parallel factor analysis, is a representation of a given tensor as a sum of rank-one components. Traditional method for accomplishing CPD is the alternating least squares (ALS) algorithm. Convergence of ALS is known to be slow, especially when some factor matrices of the tensor contain nearly collinear columns. We propose a novel variant of this technique, in which the factor matrices are partitioned into blocks, and each iteration jointly updates blocks of different factor matrices. Each partial optimization is quadratic and can be done in closed form. The algorithm alternates between different random partitionings of the matrices. As a result, a faster convergence is achieved. Another improvement can be obtained when the method is combined with the enhanced line search of Rajih et al. Complexity per iteration is between those of the ALS and the Levenberg–Marquardt (damped Gauss–Newton) method. BB 2017 3 RVO:67985556 http://hdl.handle.net/11104/0261531 S 2 Article Engineering Electrical Electronic ENGINEERING.ELECTRICAL&ELECTRONIC 2.684 0.4 5.9 0.02037 0.964 8080 0.798 Q1 2.376 320 Q2 73.144 68.5 Q2 Q2 68.511 2016 84978100769 SCOPUS 000379694800005 WOS cav_un_epca*0253212 IEEE Signal Processing Letters 1070-9908 1558-2361 Roč. 23 č. 7 2016 993 997 Institute of Electrical and Electronics Engineers