Algorithmic skeletons are polymorphic higher-order functions representing common parallelization patterns and implemented in parallel. They can be used as the building blocks of parallel and distributedapplications b...
详细信息
One of the most popular frameworks for the adaptive processing of data structures to date, was proposed by Frasconi et al. [Frasconi, P., Gori, M., & Sperduti, A. (1998). A general framework for adaptive processin...
详细信息
One of the most popular frameworks for the adaptive processing of data structures to date, was proposed by Frasconi et al. [Frasconi, P., Gori, M., & Sperduti, A. (1998). A general framework for adaptive processing of data structures. IEEE Transactions oil Neural Networks, 9(September), 768-785], who used a Backpropagation Through Structures (BPTS) algorithm [Goller, C., & Kuchler, A. (1996). Learning task-dependent distributed representations by back-propagation through structures. In Proceedings of IEEE internationalconference on neural networks (pp. 347-352);Tsoi, A. C. (1998). Adaptive processing of data structure: An expository overview and comments. Technical report in Faculty Informatics. Wollongong, Australia: University of Wollongong] to carry out supervised learning. This supervised model has been successfully applied to a number of learning tasks that involve complex symbolic structural patterns, such as image semantic structures, internet behavior, and chemical compounds. In this paper, we extend this model, using probabilistic estimates to acquire discriminative information from the learning patterns. Using this probabilistic estimation, smooth discriminant boundaries can be obtained through a process of clustering onto the observed input attributes. This approach enhances the ability of class discrimination techniques to recognize structural patterns. The proposed model is represented by a set of Gaussian Mixture Models (GMMs) at the hidden layer and a set of "weighted sum input to sigmoid function" models at the output layer. The proposed model's learning framework is divided into two phases: (a) locally unsupervised learning for estimating the parameters of the GMMs and (b) globally supervised learning for fine-tuning the GMMs' parameters and optimizing weights at the output layer. The unsupervised learning phase is formulated as a maximum likelihood problem that is solved by the expectation-maximization (EM) algorithm. The supervised learning phase
With the development of information technology of university library, the mass data of the university library has the basic characteristics of Big Data. However, the current situation of the university library is the ...
详细信息
ISBN:
(纸本)9781538621653
With the development of information technology of university library, the mass data of the university library has the basic characteristics of Big Data. However, the current situation of the university library is the lack of distributed storage and computing model for massive data, the lack of capacity to handl the diverse data sources, including the structured, semi structured and unstructured data, the lack of a simple, flexible application model of big data *** order to solve the problems in the service innovation of University Libraries in China, such as the problem of distributed storage and computation of massive data, the distributed management of diverse data sources, the simple and flexible application of big data services, this paper analyzes the research contents of big data processing, Hadoop ecosystem and the demand for big data services in University Libraries, and presents a technology framework for big data service in University Libraries based Hadoop. The framework includes the distributed storage and parallel computing model of mass data, the distributed management model of diverse data sources and the model of diversified service application for university libraries. This framework takes full account of the service innovation change of University Library under the environment of big data, such as data storage and calculation, data management and service applications et al. It can solve the key technical problems of big data service of University Library in a certain extent.
Despite the increasing popularity of shared-memory systems, there is a lack of tools for providing fault tolerance support to shared-memory applications. Check pointing is one of the most popular fault tolerance techn...
详细信息
Despite the increasing popularity of shared-memory systems, there is a lack of tools for providing fault tolerance support to shared-memory applications. Check pointing is one of the most popular fault tolerance techniques. However, check pointing cost in terms of computing time, network utilization or storage resources can be a limitation for its practical use. This work proposes different techniques for the optimization of the I/O cost in the check pointing of shared-memory parallelapplications. The proposals are extensively evaluated using the OpenMP NAS parallel Benchmarks. Results show a significant decrease of the check pointing overhead.
The amount of big data from high-throughput Next-Generation Sequencing (NGS) techniques represents various challenges such as storage, analysis and transmission of massive datasets. One solution to storage and transmi...
详细信息
ISBN:
(纸本)9781467379526
The amount of big data from high-throughput Next-Generation Sequencing (NGS) techniques represents various challenges such as storage, analysis and transmission of massive datasets. One solution to storage and transmission of data is compression using specialized compression algorithms. The existing specialized algorithms suffer from poor scalability with increasing size of the datasets and best available solutions can take hours to compress gigabytes of data. Compression and decompression using these techniques for peta-scale data sets is prohibitively expensive in terms of time and energy. In this paper we introduce paraDSRC, a parallel implementation of the DNA Sequence Reads Compression (DSRC) application using a message passing model that presents reduction of the compression time complexity by a factor of O(1/p) (where p is the number of processing units). Our experimental results show that paraDSRC achieves compression times that are 43% to 99% faster than DSRC and compression throughputs of up to 8.4GB/s on a moderate size cluster. For many of the datasets used in our experiments super-linear speedups have been registered making the implementation strongly scalable. We also show that paraDSRC is more than 25.6x faster than comparable parallel compression algorithms.
We have designed, built, and analyzed a distributedparallel storage system that will supply image streams fast enough to permit multi-user, "real-Time", video-like applications in a wide-Area ATM network-ba...
详细信息
Energy efficiency in data centres is addressed through workload management usually to reduce the operational costs and as a byproduct, the environmental footprint. This includes to minimise total power consumption or ...
详细信息
ISBN:
(纸本)9783319642031;9783319642024
Energy efficiency in data centres is addressed through workload management usually to reduce the operational costs and as a byproduct, the environmental footprint. This includes to minimise total power consumption or to minimise the power issued from non-renewable energy sources. Hence, the performance requirements of the client's applications are either totally overlooked or strictly enforced. To encourage profitable sustainability in data centres, we consider the total financial gain as a trade-off between energy efficiency and client satisfaction. We propose Carver to orchestrate energy-adaptive applications, according to performance and environmental preferences and given forecasts of the renewable energy production. We validated Carver by simulating a testbed powered by the grid and a photovoltaic array and running the Web service HP LIFE.
Circuit simulation is very important and time-consuming. parallel computing does well accelerate the calculating speeds of many applications. GPU cards have thousands of threads, so it's a good strategy to use GPU...
详细信息
ISBN:
(纸本)9781479979837
Circuit simulation is very important and time-consuming. parallel computing does well accelerate the calculating speeds of many applications. GPU cards have thousands of threads, so it's a good strategy to use GPU cards to perform circuit simulation. This paper use appropriate numerical methods, including techniques used by SPICE and Nonlinear Relaxation (relaxation-based method for solving nonlinear equations), to perform the circuit simulation. We assign the heaviest calculation portion of the simulation program to the GPU card, i.e. the nonlinear equation (derived by Newton Raphson iteration) solving portion. The complete circuit simulation program based on proposed methods has been coded and tested by solving some MOSFET circuits. The resulted speedup justifies the success of this research.
Further applications of random sampling techniques which have been used for deriving efficient parallel algorithms are presented by J. H. Reif and S. Sen [Proc. 16th internationalconference on parallelprocessing, 19...
详细信息
Further applications of random sampling techniques which have been used for deriving efficient parallel algorithms are presented by J. H. Reif and S. Sen [Proc. 16th internationalconference on parallelprocessing, 1987). This paper presents an optimal parallel randomized algorithm for computing intersection of half spaces in three dimensions. Because of well-known reductions, these methods also yield equally efficient algorithms for fundamental problems like the convex hull in three dimensions, Voronoi diagram of point sites on a plane, and Euclidean minimal spanning tree. The algorithms run in time T = O(log n) for worst-case inputs and use P = O(n) processors in a CREW PRAM model where n is the input size. They are randomized in the sense that they use a total of only polylogarithmic number of random bits and terminate in the claimed time bound with probability 1 - n(-alpha) for any fixed alpha > 0. They are also optimal in P.T product since the sequential time bound for all these problems is OMEGA(n log n). The best known deterministic parallel algorithms for two-dimensional Voronoi-diagram and three-dimensional convex hull run in O(log2 n) and O(log2 n log* n) time, respectively, while using O(n/log n) and O(n) processors, respectively.
Finite mixture models have been widely used for the modelling and analysis of data from heterogeneous populations. Maximum likelihood estimation of the parameters is typically carried out via the Expectation-Maximizat...
详细信息
ISBN:
(纸本)9781509028962
Finite mixture models have been widely used for the modelling and analysis of data from heterogeneous populations. Maximum likelihood estimation of the parameters is typically carried out via the Expectation-Maximization (EM) algorithm. The complexity of the implementation of the algorithm depends on the parametric distribution that is adopted as the component densities of the mixture model. In the case of the skew normal and skew t-distributions, for example, the E-step would involve complicated expressions that are computationally expensive to evaluate. This can become quite time-consuming for large and/or high-dimensional datasets. In this paper, we develop a multithreaded version of the EM algorithm for the fitting of finite mixture models. Due to the structure of the algorithm for these models, the E- and M-steps can be easily reformulated to be executed in parallel across multiple threads to take advantage of the processing power available in modern-day multicore machines. Our approach is simple and easy to implement, requiring only small changes to standard code. To illustrate the approach, we focus on a fairly general mixture model that includes as special or limiting cases some of the most commonly used mixture models including the normal, t-, skew normal, and skew t-mixture models. The performance gain with our approach is illustrated using two real datasets.
暂无评论