Amplitude phase shift keying (APSK) modulation signals are of low peak average power ratio and are generated to fight against the potentially non-linear high power amplifier in satellite communication. The benchmark maximum log-likelihood maximum a posteriori probability (max-log-MAP) detector requires to compute the distance squared between the received signal and each APSK constellation point for the extrinsic bit log-likelihood ratio (LLR) information. To overcome this major challenge of high detection complexity, we propose in this paper to partition and order all the APSK constellation points into rings of ordered phase shift keying (PSK) constellation points. We also propose to apply and tailor the iterative search strategies, originally used to detect multiple input multiple output (MIMO) quadrature amplitude modulation (QAM) signals, to search over the ordered irregular PSK constellation points. Compared with previously reported APSK detectors, our proposed detector achieves exactly the max-log-MAP detection, requires very low computational complexity, and is able to detect APSK signals modulated from the APSK constellation with arbitrary number of constellation points, arbitrary phase offsets, and arbitrary ring radii.

Social insurance plays crucial role for protecting the social functionality. Since different occupation would lead to different level salary and different risk, the social insurance for the people is usually designed by the characteristic of their occupation. In Taiwan, Ministry of Health and Welfare provides health care to citizens and guarantees the basic income of their elderly life. Accordingly, Ministry of Health and Welfare operates and sponsor many types of social insurances such as National Pension Insurance, Labor Pension Insurance, etc. However, due to the Population ageing, many social insurances are going to encounter the crisis of pension bankruptcy. Unfortunately, the traditional actuarial methods are based on many hypotheses to calculate cash flow, which cannot predict individuals' personal payment behavior. Besides, most traditional actuarial methods only can estimate short-term cash flow. As the result, the error of traditional actuarial methods would significantly be increased while the period of prediction target is increased. In order to predicting individuals' personal payment behavior with long-term period, we propose a Deep-Learning framework to build a prediction model. Meanwhile, the Recurrent Neural Network (RNN) architecture is adopted to effectively abstract individuals' payment behavior and precisely predict their future payment behavior with a long-term period. We conduct a comprehensive experimental study based on a real dataset collected through Taiwan's Ministry of Health and Welfare. The results demonstrate a significantly improved accuracy of our Deep Learning approach compared with the existing works.

Skyline queries are currently the most notable type of multi-criteria search algorithm. A skyline query returns all of the data points in a given a dataset that are not dominated by other data points. However, this type of query is limited by the fact that the number of results cannot be controlled. In some cases, this can result in an excessive number of results, whereas other cases result in an insufficient number of results. In this study, we propose a scheme referred to as m-representative skyline queries to provide control over the number of results that are returned. We also developed a naive algorithm and a sorted algorithm to provide additional control over the search process. Experiment results demonstrate the efficacy of the proposed approach.

Ontology is an effective formal representation of knowledge used commonly in artificial intelligence, semantic web, software engineering and information retrieval. Typically, ontologies are constructed by domain experts using domain knowledge and domain documents. However, manual acquisition of ontologies from domain documents consumes high costs. We present a support system for Vietnamese ontology construction using pattern-based mechanisms of discovering Vietnamese concepts and conceptual relations from Vietnamese text documents. As there are very few existing taxonomies constructed in Vietnamese, we use non-taxonomy based approach. The combination of association rule mining and lexical pattern based learning was used as our major method of concept extraction and conceptual relation detection. From the experiments, we show that this is a feasible solution.

Clustering is useful for mining the underlying structure of a dataset in order to support decision making since target or high-risk groups can be identified. However, for high dimensional datasets, the result of traditional clustering methods can be meaningless as clusters may only be depicted with respect to a small part of features. Taking customer datasets as an example, certain customers may correlate with their salary and education, and the others may correlate with their job and house location. If one uses all the features of a customer for clustering, these local-correlated clusters may not be revealed. In addition, processing high dimensions and large datasets is a challenging problem in decision making. Searching all the combinations of every feature with every record to extract local-correlated clusters is infeasible, which is in exponential scale in terms of data dimensionality and cardinality. In this paper, we propose a scalable 2-Leveled Approximated Hyper-Image-based Clustering framework, referred as 2L-HIC-A, for mining local-correlated clusters, where each level clustering process requires only one scan of the original dataset. Moreover, the data-processing time of 2L-HIC-A can be independent of the input data size. In 2L-HIC-A, various well-developed image processing techniques can be exploited for mining clusters. In stead of proposing a new clustering algorithm, our framework can accommodate other clustering methods for mining local-corrected clusters, and to shed new light on the existing clustering techniques.

The processes of software engineering and the industries themselves tend to be fluctuant. There are numerous factors, such as software methodologies, technologies, supporting tools, and process managements, which may significantly affect the strategies and activities of a software development process. We improve the PIE approach and propose an XML-based meta-model for process and agent-based integrated software development environment (PRAISE). PRAISE includes both the external representation in UML and its internal representation in XML, and can be used to support the integration of software development in a global aspect.

Abstract Many disparate software analysis/design methods and tools (called paradigms in this paper) individually offer faster and more efficient software design/evolution, yet they are generally partially incompatible. Therefore, when they are used together, they suffer from a lack of communication and integration. Unanticipated and inadvertent inconsistencies arising from the implicit incompatibilities result in significantly increased costs and errors during software maintenance. To improve both software evolution and maintenance, we propose an XML‐based model that effectively unifies and integrates current software paradigms while helping reduce ripple effects and related software maintenance difficulties. Copyright © 2003 John Wiley & Sons, Ltd.

In this paper we present an efficient data compression scheme, the template run-length encoding (TRLE) scheme, for image composition of parallel volume rendering systems. Given an image with 2n/spl times/2n pixels, in the TRLE scheme, the image is treated as n/spl times/n blocks and each block has 2/spl times/2 pixels. Since a pixel can be a blank or non-blank pixel, there are 16 templates in a block. To compress an image, the TRLE scheme uses the templates to encode blocks row by row. Blocks in the same row are encoded as a TRLE-sequence. By packing all TRLE-sequences in a packet, the packet is the compressed partial image that can be sent/received among processors. To evaluate the performance of the TRLE scheme, we compare the proposed scheme with the BR, the RLE, and the BRLC schemes. Since a data compression scheme needs to cooperate with some data communication schemes, in the implementation, the binary-swap (BS), the parallel-pipelined (PP), and the rotate-tiling (RT) data communication schemes are used. By combining the four data compression schemes with the three data communication schemes, we have twelve image composition methods. These twelve methods are implemented on a PC cluster The data computation time and the data communication time are measured. The experimental results show that the TRLE data compression scheme with the RT data communication scheme outperforms other eleven image composition methods.

The marching cubes method is a well-known surface extraction method by using the surface configurations of cubes for surface rendering of volume data. The marching cubes method has three main disadvantages: it is time consuming, ambiguous, and holes are generated. All these disadvantages come from the use of the surface configurations of cubes. We propose an efficient surface extraction method, the marching voxels method, for surface rendering of volume data. Instead of using the surface configurations of cubes, the marching voxels method first generates triangles for inner voxels. Then it combines the triangles of inner voxels to produce the surface of an object. Finally, the surface of an object is projected to a plane to form the final image. Since the marching voxels method considers the combination of triangles of voxels not cubes and the combination of triangles is performed in a deterministic way, there is neither ambiguous case of a combination nor holes for the generated surface. The experimental results show that the marching voxels method saves about 30% of the surface rendering time compared to the marching cubes method for test samples.

The association rule mining can be divided into two steps. The first step is to find out all frequent itemsets, whose occurrences are greater than or equal to the user-specified threshold. The second step is to generate reliable association rules based on all frequent itemsets found in the first step. Identifying all frequent itemsets in a large database dominates the overall performance in the association rule mining. In this paper, we propose an efficient hash-based method, HMFS, for discovering the maximal frequent itemsets. The HMFS method combines the advantages of both the DHP (Direct Hashing and Pruning) and the Pincer-Search algorithms. The combination leads to two advantages. First, the HMFS method, in general, can reduce the number of database scans. Second, the HMFS can filter the infrequent candidate itemsets and can use the filtered itemsets to find the maximal frequent itemsets. These two advantages can reduce the overall computing time of finding the maximal frequent itemsets. In addition, the HMFS method also provides an efficient mechanism to construct the maximal frequent candidate itemsets to reduce the search space. We have implemented the HMFS method along with the DHP and the Pincer-Search algorithms on a Pentium III 800 MHz PC. The experimental results show that the HMFS method has better performance than the DHP and the Pincer-Search algorithms for most of test cases. In particular, our method has significant improvement over the DHP and the Pincer-Search algorithms when the size of a database is large and the length of the longest itemset is relatively long.

The Fuzzy C-Means (FCM) algorithm is commonly used for clustering. The performance of the FCM algorithm depends on the selection of the initial cluster center and/or the initial membership value. If a good initial cluster center that is close to the actual final cluster center can be found, the FCM algorithm will converge very quickly and the processing time can be drastically reduced. The authors propose a novel algorithm for efficient clustering. This algorithm is a modified FCM called the psFCM algorithm, which significantly reduces the computation time required to partition a dataset into desired clusters. We find the actual cluster center by using a simplified set of the original complete dataset. It refines the initial value of the FCM algorithm to speed up the convergence time. Our experiments show that the proposed psFCM algorithm is on average four times faster than the original FCM algorithm. We also demonstrate that the quality of the proposed psFCM algorithm is the same as the FCM algorithm.

In today's cyber-world, network faults not only bring unexpected damages and misgiving to the enterprises, but also cause the loss of customers without a sign. To promote the quality of network services and the competition of enterprises, plenty of network management activities are funded to develop competent fault management mechanisms. In this paper, we offer a fault diagnosis mechanism for effective and automated network fault isolation. This mechanism uses finite state machine (FSM) to model refined fault propagation behaviors. The concept called fault propagation duration is also incorporated into the mechanism and combined with the FSM-based fault propagation knowledge to realize the mechanism. Based on the diagnosis mechanism, an automated fault diagnosis system called alarm correlation view (ACView) for isolating network faults is built. This diagnosis system can not only provide the automated process of alarm collection, alarm correlation, and fault isolation, but identify the propagation of each fault with a corresponding severity.

The binary-swap and the parallel-pipelined methods are two popular image composition methods for volume rendering on distributed memory multicomputers. However, these methods either restrict the number of processors to a power of two or require many steps to transform image data that results in high communication overheads. In this paper, we present an efficient image composition method, the rotate-tiling (RT), for parallel volume rendering on distributed memory multicomputers. The RT method can fully utilize all available processors and minimize the communication overheads. In addition, we provide data compression method, the template run-length encoding (TRLE), to further reduce the communication data size. To evaluate the performance of the RT method, we compare the proposed method with the binary-swap method and the parallel-pipelined method. Both theoretical analysis and experimental test are conducted. In the theoretical analysis, we analyze the best performance bound of the RT method in terms of the startup time, the data transmission time, the number of processors, and the number of initial block of a sub-image. In the experimental test, we have implemented these three methods on an SP2 parallel machine. Three volume datasets are used as test samples. The experimental results show that our method outperforms the binary-swap and the parallel-pipelined methods for all test samples and match the results analyzed in the theoretical analysis. For the TRLE method, the experimental results show that the TRLE method can further reduce the composition time for these three methods.

In many scientific applications, array redistribution is usually required to enhance data locality and reduce remote memory access in many parallel programs on distributed memory multicomputers. Since the redistribution is performed at runtime, there is a performance trade-off between the efficiency of the new data decomposition for a subsequent phase of an algorithm and the cost of redistributing data among processors. In this paper, we present a generalized processor mapping technique to minimize the amount of data exchange for BLOCK-CYCLIC(kr) to BLOCK-CYCLIC(r) array redistribution and vice versa. The main idea of the generalized processor mapping technique is first to develop mapping functions for computing a new rank of each destination processor. Based on the mapping functions, a new logical sequence of destination processors can be derived. The new logical processor sequence is then used to minimize the amount of data exchange in a redistribution. The generalized processor mapping technique can handle array redistribution with arbitrary source and destination processor sets and can be applied to multidimensional array redistribution. We present a theoretical model to analyze the performance improvement of the generalized processor mapping technique. To evaluate the performance of the proposed technique, we have implemented the generalized processor mapping technique on an IBM SP2 parallel machine. The experimental results show that the generalized processor mapping technique can provide performance improvement over a wide range of redistribution problems.

In the sort-last-sparse parallel volume rendering system on distributed memory multicomputers, as the number of processors increases, in the rendering phase, we can get a good speedup because each processor renders images locally without communicating with other processors. However, in the compositing phase, a processor has to exchange local images with other processors. When the number of processors is over a threshold, the image compositing time becomes a bottleneck. In this paper, we proposed three compositing methods, the binary-swap with bounding rectangle method, the binary-swap with run-length encoding and static load-balancing method, and the binary-swap with bounding rectangle and run-length encoding method, to efficiently reduce the compositing time in the sort-last-sparse parallel volume rendering system on distributed memory multicomputers. The proposed methods were implemented on an SP2 parallel machine along with the binary-swap compositing method. The experimental results show that the binary-swap with bounding rectangle and run-length encoding method has the best performance among the four methods.