Vol. 36 · No. 3

The N-Person Iterated Prisoner's Dilemma (NIPD) is an interesting game that has proved to be very useful to explore the emergence of cooperation in multi-player scenarios. Within this game, the way that agents are interconnected is a key element that influences cooperation. In this context, complex networks provide a realistic model of the topological features found in Nature and in many social and technological networks. Considering these networks, it is interesting to study the network evolution, given the possibility that agents can change their neighbors (dynamic rewire), when non-cooperative behaviors are detected. In this paper, we present a model of the NIPD game where a population of genetically-coded agents compete altogether. We analyze how different game parameters, and the network topology, affect the emergence of cooperation in static complex networks. Based on that, we present the main contribution of the paper that concerns the influence of dynamic rewiring in the emergence of cooperation over the NIPD.

Finding trustworthy worker is a longstanding issue in crowdsourcing systems. On traditional crowdsouring platforms, like Amazon Mechanical Turk:Amazon Mechanical Turk, the trustworthiness of a worker is usually based on the contextual information, like different types of tasks and different reward amounts of tasks. However, with the combination of OSNs and the crowdsourcing applications in social crowd, in addition to the above mentioned task based contexts, the social contexts like the social relationships and the social positions of participants can greatly assist requestors to select trustworthy workers. In order to select the trustworthy workers in social crowd, in this paper, we first present a contextual social network structure which contains complex social contexts. Then we propose a trust evaluation model taking both contexts information and the requirements of requestors into consideration, which leads to the trust worker selection in social crowd as a classical NP-Complete Multi-Constrained Optimal Path (MCOP) selection problem. For solving this challenging problem, we propose a new efficient and effective approximation Context-Aware Trust-Oriented Worker Selection algorithm CAT. The results of our experiments conducted on four real OSN datasets illustrate the superiority of our method in trustworthy worker selection.

Infrastructure management is of key importance in a wide array of computer and network environments. The use of virtualization in cloud datacenters has driven the communications and computing convergence to a common operational entity. Failure to effectively manage the involved infrastructure results as impediments in provisioning a successful service. Information models facilitate the infrastructure management and current solutions can be effectively applied in most datacenter scenarios, apart from cases where the networking architecture relies heavily on systems virtualization. In this paper we propose an information model for managing virtual network architectures, where hypervisors and computing server resources are deployed as the basis of the networking layer. We provide a successful proof of concept by managing a virtual machine-based network infrastructure acting as an IP routing platform using statistical methods. Our proposal enables a dynamic reconfiguration of allocated infrastructure resources adapting, in real-time, to variations in the imposed workload.

Data locality improvement and nested loops parallelization are two complementary and competing approaches for optimizing loop nests that constitute a large portion of computation times in scientific and engineering programs. While there are effective methods for each one of these, prior studies have paid less attention to address these two simultaneously. This paper proposes a unified approach that integrates these two techniques to obtain an appropriate locality conscious loop transformation to partition the loop iteration space into outer parallel tiled loops. The approach is based on the polyhedral model to achieve a multidimensional affine scheduling as a transformation that result the largest groups of tilable loops with maximum coarse grain parallelism, as far as possible. Furthermore, tiles will be scheduled on processor cores to exploit maximum data reuse through scheduling tiles with high volume of data sharing on the same core consecutively or on different cores with shared cache at around the same time.

In the last decade, the processing of the high dimensional data became inevitable task in many areas of research and daily life. Feature selection (FS), as part of the data processing methodology, is an important step in knowledge discovery. This paper proposes nine variation of two-step feature selection approach with filter FS employed in the first step and exhaustive search in the second step. The performance of the proposed methods is comparatively analysed from the stability and predictive performance point of view. As the obtained results indicate the choice of the filter FS in the first stage has strong influence on the resulting stability. Here, the choice of univariate Pearson correlation coefficient based FS method appears to provide the most stable results.

Multiple kernel learning (MKL) has recently received significant attention due to the fact that it is able to automatically fuse information embedded in multiple base kernels and then find a new kernel for classification or regression. In this paper, we propose a coupled multiple kernel learning method for supervised classification (CMKL-C), which comprehensively involves the intra-coupling within each kernel, inter-coupling among different kernels and coupling between target labels and real ones in MKL. Specifically, the intra-coupling controls the class distribution in a kernel space, the inter-coupling captures the co-information of base kernel matrices, and the last type of coupling determines whether the new learned kernel can make a correct decision. Furthermore, we deduce the analytical solutions to solve the CMKL-C optimization problem for highly efficient learning. Experimental results over eight UCI data sets and three bioinformatics data sets demonstrate the superior performance of CMKL-C in terms of the classification accuracy.

This paper presents a new approach of dynamic clustering based on improved Particle Swarm Optimization (PSO) and which is applied to image segmentation (called DCPSONS). Firstly, the original PSO algorithm is improved by using diversity mechanism and neighborhood search strategy. The improved PSO is then combined with the well-known data clustering k-means algorithm for dynamic clustering problem where the number of clusters has not yet been known. Finally, DCPSONS is applied to image segmentation problem, in which the number of clusters is automatically determined. Experimental results in using sixteen benchmark data sets and several images of synthetic and natural benchmark data demonstrate that the proposed DCPSONS algorithm substantially outperforms other competitive algorithms in terms of accuracy and convergence rate.

This manuscript deals with optimizing the numerical method called the method of moments (MoM). This method is widely utilized for field computation of 3D structures. MoM is exploited in hydraulics as well as in the electromagnetic field theory. Emphasis is put on minimizing calculations necessary for constructing a system of linear equations exploiting symmetry or similarity of elements of geometric structure. The manuscript also contains a comparison of computing times using standard MoM and a proposed modified MoM keeping the same structure of the solved construction.

String similarity join is a basic and essential operation in many applications. In this paper, we investigate the problem of string similarity join with edit distance constraints. A trie-based edit similarity join framework has been proposed recently. The main advantage of existing trie-based algorithms is support for similarity join on short strings. The main problem is when joining long and distant strings. These methods generate and maintain lots of similar prefixes called active nodes which need to be further removed in a subsequent pruning phase. With large edit distance, the number of active nodes becomes quite large. In this paper, we propose a new trie-based join algorithm called PreJoin, which improves upon current trie-based join methods. It efficiently finds all similar string pairs using a novel active-node generation method, which minimizes the number of generated active nodes by applying the pruning heuristics early in the generation process. The performance of PreJoin is scaled in two different ways: First, a dynamic reordering of the trie index is used to accelerate the search for similar string pairs. Second, a partitioning method of string space is used to improve performance on large edit distance thresholds. Experiments show that our approach is highly efficient for processing short as well as long strings, and outperforms the state-of-the-art trie-based join approaches by a factor five.

Regression testing, which can improve the quality of software systems, is a useful but time consuming method. Many techniques have been introduced to reduce the time cost of regression testing. Among these techniques, test case prioritization is an effective technique which can reduce the time cost by processing relatively more important test cases at an earlier stage. Previous works have demonstrated that some greedy algorithms are effective for regression test case prioritization. Those algorithms, however, have lower stability and scalability. For this reason, this paper proposes a new regression test case prioritization approach based on the improved Annealing-Genetic algorithm which incorporates Simulated Annealing algorithm and Genetic algorithm to explore a bigger potential solution space for the global optimum. Three Java programs and five C programs were employed to evaluate the performance of the new approach with five former approaches such as Greedy, Additional Greedy, GA, etc. The experimental results showed that the proposed approach has relatively better performance as well as higher stability and scalability than those former approaches.