Matrix tri-factorization over the tropical semiring

Tropical semiring has proven successful in several research areas, including optimal control, bioinformatics, discrete event systems, or solving a decision problem. In previous studies, matrix two-factorization algorithm based on the tropical been applied to investigate bipartite and tripartite networks. Tri-factorization algorithms standard linear algebra are used for tasks such as data fusion, co-clustering, completion, community detection, more. However, there is currently no tri-factorization approach, which would allow analysis of multipartite networks with high number parts. To address this, we propose triFastSTMF algorithm, performs over semiring. We apply it analyze four-partition network structure recover edge lengths network. show that similarly Fast-NMTF terms approximation prediction performance when fitted whole When trained specific subnetwork predict network, outperforms by orders magnitude smaller error. The robustness due operations, less prone large values compared operations.