Parallel tensor factorization for relational learning
Source of Publication
Neural Computing and Applications
Link prediction is a statistical relational learning problem that has a variety of applications in recommender systems, expert systems, and knowledge bases. Numerous approaches have already been devised to solve the problem. Tensor factorization is one of the ways to solve the link prediction problem. Many tensor factorization techniques have been devised in the last few decades, including Tucker, CANDECOMP/PARAFAC, and DEDICOM. RESCAL is one of the famous tensor factorization technique that can solve large scale problems with relatively less time and space complexity. The time complexity of RESCAL can further be reduced by making it parallel. This variant can also be applied to large scale datasets. This article focuses on devising a parallel version for RESCAL. A decent decrease in execution time has been observed in the execution of parallel RESCAL.
Springer Science and Business Media Deutschland GmbH
Expert systems, Factorization, Large dataset, Knowledge basis, Large-scale datasets, Large-scale problem, Parallel version, Relational learning, Statistical relational learning, Tensor factorization, Time and space complexity, Tensors
Al-Obeidat, Feras; Rocha, Álvaro; Khan, Muhammad Shahrose; Maqbool, Fahad; and Razzaq, Saad, "Parallel tensor factorization for relational learning" (2021). All Works. 2628.
Indexed in Scopus