Gtensor

Abstract

Given a large tensor, how can we analyze it efficiently? Multidimensional arrays or tensors have been widely used to model real-world data. Tensor decomposition plays an important role in analyzing trends and major factors in tensors. While several tensor analysis tools have been developed, they show slow running time and limited scalability due to their heavy computational requirements. In this paper, we propose Gtensor, a fast and accurate tensor mining library using GPUs. Gtensor is carefully designed to utilize GPUs in addition to CPUs for fast analysis of tensors. Gtensor provides versatile tensor algebra tools including tensor decomposition, tensor generation, tensor-tensor operation, and tensor manipulation. We invite our audience to interact with Gtensor and analyze real-world tensor data in the domains of movie recommendation, place recommendations, and academic citation. CCS CONCEPTS • Computing methodologies → Massively parallel algorithms; • Information systems → Data mining.