Vol. 32 · No. 4

Bootstrapped aggregation of classifiers, also referred to as bagging, is a classic meta-classification algorithm. We extend it to a two-stage architecture consisting of an initial voting amongst one-versus-all classifiers or single-class recognizers, and a second stage of one-versus-one classifiers or two-class discriminators used for disambiguation. Since our method constructs an ensemble of elementary classifiers, it lends itself very well to parallelization. We describe a static workload balancing strategy for embarrassingly parallel classifier construction as well as a parallelization of the classification process with the message passing interface. In our experiments, which are evaluated in terms of classification performance and speed-up, we obtained an up to three-fold increase in precision and significantly increased recall values.

In this paper, we propose QMBRi, the inverse representation of the quantized minimum bounding rectangles (MBRs) scheme, which compresses a minimum bounding rectangle key into one byte for spatial-data compression. QMBRi is a novel spatial-data compression scheme that is based on inverse quantization and overcomes the shortcomings of conventional relative coordination or quantization schemes. If a spatial data is far from the starting point of the search region, the relative coordination scheme does not guarantee compression. In a quantization scheme, since the MBRs are expanded, the overlapping of MBRs is increased and the search performance is reduced. The proposed scheme overcomes these shortcomings, and simulation results suggest that it performs better than other schemes.

This research presents an optimization technique for multiple routes generation using simulated niche based particle swarm optimization for dynamic online route planning, optimization of the routes and proved to be an effective technique. It effectively deals with route planning in dynamic and unknown environments cluttered with obstacles and objects. A simulated niche based particle swarm optimization (SN-PSO) is proposed using modified particle swarm optimization algorithm for dealing with online route planning and is tested for randomly generated environments, obstacle ratio, grid sizes, and complex environments. The conventional techniques perform well in simple and less cluttered environments while their performance degrades with large and complex environments. The SN-PSO generates and optimizes multiple routes in complex and large environments with constraints. The traditional route optimization techniques focus on good solutions only and do not exploit the solution space completely. The SN-PSO is proved to be an efficient technique for providing safe, short, and feasible routes under dynamic constraints. The efficiency of the SN-PSO is tested in a mine field simulation with different environment configurations and successfully generates multiple feasible routes.

As cloud computing becomes prevalent, more and more sensitive information are being centralized into the cloud. For the protection of data privacy, sensitive data usually have to be encrypted before outsourcing, which makes effective data utilization a very challenging task. In this paper, we propose a new method to enable effective fuzzy keyword search in a multi-user system over encrypted cloud data while maintaining keyword privacy. In this new system, differential searching privileges are supported, which is achieved with the technique of attribute-based encryption. Edit distance is utilized to quantify keywords similarity and develop fuzzy keyword search technique, which achieve optimized storage and representation overheads. We further propose a symbol-based trie-traverse searching scheme to improve the search efficiency. Through rigorous security analysis, we show that our proposed solution is secure and privacy-preserving, while correctly realizing the goal of fuzzy keyword search with multiple users.

This paper describes our generic framework for detecting termination of programs handling infinite and complex data domains, such as pointer structures. The framework is based on a counterexample-driven abstraction refinement loop. We have instantiated the framework for programs handling tree-like data structures, which allowed us to prove automatically termination of programs such as the depth-first tree traversal, the Deutsch-Schorr-Waite tree traversal, or the linking leaves algorithm.

Constrained financial portfolio optimization is a challenging domain where the use of multiobjective evolutionary algorithms has been thriving over the last few years. One of the major issues related to this problem is the dependence of the results on a set of parameters. Given the nature of financial prediction, these figures are often inaccurate, which results in unreliable estimates for the efficient frontier. In this paper we introduce a resampling mechanism that deals with uncertainty in the parameters and results in efficient frontiers that are more robust. We test this idea on real data using four multiobjective optimization algorithms (NSGA-II, GDE3, SMPSO and SPEA2). The results show that resampling significantly increases the reliability of the resulting portfolios.

Software architecture plays an important role in the success of modern, large and distributed software systems. For many of the software systems – especially safety-critical ones – it is important to specify their architectures using formal modeling notations. In this case, it is possible to assess different functional and non-functional properties on the designed models. Graph Transformation System (GTS) is a formal yet understandable language which is suitable for architectural modeling. Most of the existing works done on architectural modeling and analysis by GTS are concentrated on functional aspects, while for many systems it is crucial to consider non-functional aspects for modeling and analysis at the architectural level. In this paper, we present an approach to performance analysis of software architectures specified through GTS. To do so, we first enrich the existing architectural style – specified through GTS – with performance information. Then, the performance models are generated in PEPA (Performance Evaluation Process Algebra) – a formal language based on the stochastic process algebra – using the enriched GTS models. Finally, we analyze different features like throughput, utilization of different software components, etc. on the generated performance models. All the main concepts are illustrated through a case study.

One of the major challenges in software engineering research is to manage software artifacts effectively. However, software artifacts are often changed during software development, the full, one-time integration technique is not feasible to manage such heterogeneity and evolving data. In this paper, we concern about the application of dataspace techniques, which emphasize the idea of pay-as-you-go data management, to software artifacts management. To this end, we present a loosely structured data model based on the current dataspace models to describe software artifacts, and a strategy to query this model. We also present how to gradually add semantics to query processing for improving the precision and recall of query results. Furthermore, the validity of our work is proved by experiment. Finally, the differences between our work and traditional work are discussed.

Melt volume-flow rate (MVR) is one of the most important quality indicators of composite materials, which depends on the proportion of the component materials. This paper reports the development of a low complexity fuzzy model that describes the relation between percentage amount of multiwall carbon nanotube (MWCNT), acrylonitrile-butadiene-styrene (ABS), polycarbonate (PC) and MVR of the resulting composite. The rule base was generated from a sample data set obtained from experiments by the rule base extension using default set shapes (RBE-DSS) method, and the applied fuzzy inference technique was the least squares method based fuzzy rule interpolation (LESFRI). The resulting model was validated against a separate test data set as well, and it was compared to a fuzzy model generated by a traditional commercial software tool.

The primary goal of the presented experiment was to judge the usefulness of FPGA technology in the sorting operation performed by computer systems. We were interested to see if it was possible to achieve better system performance and lower energy consumption when the CPU is supported by FPGA chips. The method of custom processing was applied. We proposed dedicated sorting hardware to increase performance and save energy. Our concept addresses High Throughput Computing (HTC) systems. The custom hardware approach was proposed because this technique is available in supercomputing infrastructures today. Another important issue of the work is that the hardware was programmed using High Level Language (HLL). As a semiconductor platform for hardware implementation the FPGA was chosen. We evaluated the efficiency of an FPGA based sorting processor that was programmed in Mitrion-C HLL. The FPGA approach was compared to the CPU approach in terms of efficiency and power consumption.

In this paper we propose a novel kernel for text categorization. This kernel is an inner product defined in the feature space generated by all word combinations of specified length. A word combination is a collection of unique words co-occurring in the same sentence. The word combination of length k is weighted by the k rm th root of the product of the inverse document frequencies (IDF) of its words. By discarding word order, the word combination features are more compatible with the flexibility of natural language and the feature dimensions of documents can be reduced significantly to improve the sparseness of feature representations. By restricting the words to the same sentence and considering multi-word combinations, the word combination features can capture similarity at a more specific level than single words. A computationally simple and efficient algorithm was proposed to calculate this kernel. We conducted a series of experiments on the Reuters-21578 and 20 Newsgroups datasets. This kernel achieves better performance than the word kernel and word-sequence kernel. We also evaluated the computing efficiency of this kernel and observed the impact of the word combination length on performance.