Исследование эффективности алгоритмов факторизации матриц на GPU-устройствах для обработки данных рекомендательных систем

Владимир Сердюк, Санкт-Петербургский политехнический университет Петра Великого. Институт компьютерных наук и технологий, Ольга Колосова, Александр Щукин
2019
The object of the research is the matrix factorization algorithms on GPU-devices for recommender systems data processing. The subject of the research is the performance of matrix factorization algorithms on GPU-devices for recommender systems data processing. The goal is to study the efficiency of matrix factorization algorithms on GPU devices for recommender systems data processing. To achieve this goal following tasks are solved: an overview of matrix factorization algorithms in the context
more » ... ms in the context of their application in recommender systems; overview of the architecture and programming tools of modern graphics accelerators; implementation of the selected algorithm using graphics accelerators and optimization taking into account the architectural features of modern graphics accelerators; experimental study of the effectiveness of the proposed optimization; comparing the performance of an optimized implementation with known alternatives using graphics accelerators.
doi:10.18720/spbpu/3/2019/vr/vr19-2203 fatcat:djhyn7u42ngffnq752nuq4z6pm