Bilkent University
Department of Computer Engineering


Parallel Tensor Decomposition on Spark


Selim Eren Bekçe
MS Student
Computer Engineering Department
Bilkent University

Tensors are higher order matrices that have wide variety of usage and their decomposition is used to discover latent features from multi-dimensional data. There are existing serial tools for decomposition but large size tensors are unlikely to fit in one machine's memory or computing power, therefore parallel algorithms need to be employed. In this presentation, we will explore PARAFAC (CP) decomposition technique on Apache Spark, which is a popular parallelization framework with its revolutionary programming paradigm called Resilient Distributed Datasets.


DATE: 03 April, 2017, Monday @ 15:40