GPUMLib: An Efficient Open-Source GPU Machine Learning Library

Graphics Processing Units (GPUs) placed at our disposal an unprecedented computational-power, largely surpassing the performance of cutting-edge CPUs (Central Processing Units). The high-parallelism inherent to the GPU makes this device especially well-suited to address Machine Learning (ML) problems with prohibitively computational intensive tasks. Nevertheless, few ML algorithms have been implemented on the GPU and most are not openly shared, posing difficulties for researchers and engineers aiming to develop GPU-based systems. To mitigate this problem, we propose the creation of an open source GPU Machine Learning Library (GPUMLib) that aims to provide the building blocks for the development of efficient GPU ML software. Experimental results on benchmark datasets show that the algorithms already implemented yield significant time savings over the CPU counterparts.