Performance improvement of data mining in Weka through multi-core and GPU acceleration: opportunities and pitfalls

Abstract

International audienceData mining tools may be computationally demanding, which leads to an increasing interest on par- allel computing strategies in order to improve their per- formance. While multi-core processors and Graphics Processing Units (GPUs) accelerators increased the com- puting power of current desktop computers, we observe that desktop-based data mining tools do not take full advantage of these architectures yet. This paper investi- gates strategies to improve the performance of Weka, a popular data mining tool, through multi-core and GPU acceleration. Using performance profiling of Weka, we identify operations that could improve the data mining performance when parallelized. We selected two of these operations, and analyze the impact of their parallel exe- cution on Weka’s performance. These experiments demonstrate that while significant speedups can be achieved, all operations are not prone to be parallelized, which reinforces the need for a careful and well-studied selection of the candidates

Dependency Prediction of Long-Time Resource Uses in HPC Environment

“…• Performance Comparison: The table-9 shows a specific comparison between the proposed method and other major resource prediction methods like (the regressive ensemble approach for predicting resource (REAP [23]), Bayesian Approach (BN [24]), cost-aware auto-scaling approach (LR [25]), the advanced model for efficient workload prediction (RF [26]), and fuzzy neural network (FNN [20]). For all our experiments we have used Root mean square error and absolute root error at (λ = 0.5)%.…”

“…They uncovered a good speedup on a large space production. A remarkable finding about the performance of hashing algorithm is presented by Tiago et al, [23]. Even after having these methods to distribute data-mining equipment in a cluster or grid, no better technique was made to effectively use them on any single machine.…”

Weka, áreas de aplicación y sus algoritmos: una revisión sistemática de literatura

“…Además, normalmente los algoritmos proporcionan parámetros que cuando se ajustan pueden mejorar los resultados (por ejemplo, precisión). Sin embargo, a menudo la precisión y el tiempo de cálculo están correlacionados y cuando el primero se requiere alto, el segundo también se ve afectado (Engel, Charão, Kirsch-Pinheiro, & Steffenel, 2015).…”