Showing papers presented at "Parallel and Distributed Processing Techniques and Applications in 2015"

Crowdsourcing the cloud: energy-aware computational offloading for pervasive community-based cloud computing

Abstract: Adaptive offloading systems achieve context specific optimization on mobile and pervasive devices by offloading computational components to a resource copious remote server or cloud. However, with the recent advancement in computational capacity of mobile and pervasive devices, adaptive offloading could facilitate the formation of ad-hoc cloud-like environments using collections of mobile and pervasive devices, with reduced reliance on centralized infrastructure. Therefore, in this paper, we formulate a decision-making strategy for global adaptive offloading that distributes application components to community-based clouds formed from multiple collaborating peers. The goal was to extend the collaboration and application lifetime by optimizing the Time to Failure (TTF) of devices due to energy depletion, while meeting application specific performance constraints. Specifically, a max-min technique was used to maximise the minimum TTF in order to balance energy consumption across collaborating devices. The efficacy, performance and scalability of the formulated model were evaluated with the proposed algorithm producing an optimal solution to the specified model, using integer linear programming, in affordable time and energy for a range of application and collaboration sizes.

Empirical Study of Time Efficiency and Accuracy of Support Vector Machines Using an Improved Version of PSVM

Abstract: We present a significantly improved implementation of a parallel SVM algorithm (PSVM) together with a comprehensive experimental study. Support Vector Machines (SVM) is one of the most well-known machine learning classification techniques. PSVM employs the Interior Point Method, which is a solver used for SVM problems that has a high potential of parallelism. We improve PSVM regarding its structure and memory management for contemporary processor architectures. We perform a number of experiments and study the impact of the reduced column size p and other important parameters as C and gamma on the class-prediction accuracy and training time. The experimental results show that there exists a threshold between the number of computational cores and the training time, and that choosing an appropriate value of p effects the choice of the C and gamma parameters as well as the accuracy.