Applied Mechanics and Materials Vols. 284-287

Abstract: Academic search engines, such as Google Scholar and Scirus, provide a Web-based interface to effectively find relevant scientific articles to researchers. However, current academic search engines are lacking the ability to cluster the search results into a hierarchical tree structure. In this paper, we develop a post-search academic search engine by using a mixed clustering method. In this method, we first adopt a suffix tree clustering and a two-way hash mechanism to generate all meaningful labels. We then develop a divisive hierarchical clustering algorithm to organize the labels into a hierarchical tree. According to the results of experiments, we conclude that using our mixed clustering method to cluster the search results can give significant performance gains than current academic search engines. In this paper, we make two contributions. First, we present a high performance academic search engine based on our mixed clustering method. Second, we develop a divisive hierarchical clustering algorithm to organize all returned search results into a hierarchical tree structure.

Abstract: We have explored an approach for building a multi-classifier system in a GA-based inductive learning environment. In our approach, multiple base classifiers are combined to build a multi-classifier system. A base classifier consists of a general classifier and a meta-classifier. The role of a general classifier is to perform regular classification task and that of a meta-classifier is to evaluate the classification result of its general classifier and decide whether the base classifier participates into a final decision-making process or not. The paper discusses our approach in details and presents some empirical results that show the improvement we can achieve with our approach.

Abstract: In a noise environment probabilistic fuzzy clustering will force the noise into one or more clusters, seriously influencing the main dataset structure. We extend Type-1 membership values to Type-2 by assigning a possibilistic-membership function to each Type-1 membership value. The idea in building the Type-2 fuzzy sets is based simply on the fact that, for the same Type-1 membership value, the secondary membership function should make the larger possibility value greater than the smaller possibility value. This paper presents an efficient combined probabilistic and possibilistic method for building Type-2 fuzzy sets. Utilizing this concept we present a Type-2 FCM (T2FCM) that is an extension of the conventional FCM. The experimental results show that the T2FCM is less susceptible to noise than the Type-1 FCM. The T2FCM can ignore the inlier and outlier interrupt. The clustering results show the robustness of the proposed T2FCM because a reasonable amount of noise data does not affect its clustering performance.

Abstract: In this study, we report a voting behavior analysis intelligent system based on data mining technology. From previous literature, we have witnessed increasing number of studies applied information technology to facilitate voting behavior analysis. In this study, we built a likely voter identification model through the use of data mining technology, the classification algorithm used here constructs decision tree model to identify voters and non voters. This model is evaluated by its accuracy and number of attributes used to correctly identify likely voter. Our goal is to try to use just a small number of survey questions while maintaining the accuracy rates of other similar models. This model was built and tested on Taiwan’s Election and Democratization Study (TEDS) data sets. According to the experimental results, the proposed model can improve likely voter identification rate and this finding is consistent with previous studies based on American National Election Studies.

Abstract: Visually convincing content-aware image resizing, which preserves semantically important image content, has been actively researched in recent years. This paper proposes a resizing detector that reveals the trace of seam carving and seam insertion. To unveil the evidence of seam carving, we exploit energy bias of seam carved images. In addition, the correlation between adjacent pixels is analyzed to estimate the inserted seams. Empirical evidence from a large database of test images demonstrates the superior performance of the proposed detector under a variety of settings.

Abstract: A novel VLSI architecture for kernel fuzzy c-means algorithm is presented in this paper. The architecture consists of efficient circuits for the computation of kernel functions, membership coefficients and cluster centers. In addition, the usual iterative operations for updating the membership matrix and cluster centers are merged into one single updating process to evade the large storage requirement. The circuit is used as a hardware accelerator of a softcore processor in a system-on-programmable chip for physical performance measurement. Experimental results show that the proposed solution is an effective alternative for cluster analysis with low computational cost and high performance.

Abstract: This paper applies the enhanced particle swarm optimization (EPSO) algorithm to solve the unit commitment problem, and compares the results obtained against previous work. EPSO can improve the search quality and also generate a better result through optimization, because ants produced randomly by the pheromone process are not necessary better. The proposed model uses combined carbon finance and spot market formulation, and help energy produces decide when these commitments could be beneficial.

Abstract: Estimation of distribution algorithms (EDAs) constitute a new branch of evolutionary optimization algorithms, providing effective and efficient optimization performance in a variety of research areas. Recent studies have proposed new EDAs that employ mutation operators in standard EDAs to increase the population diversity. We present a new mutation operator, a matrix transpose, specifically designed for Bayesian structure learning, and we evaluate its performance in Bayesian structure learning. The results indicate that EDAs with transpose mutation give markedly better performance than conventional EDAs.

Abstract: The goal of this study is to explore the factor effect of learning vector quantization. The manipulated factors are training pattern, learning rate, types of mixed data, and hidden node. The results showed that the average accuracy for severe overlap data was significantly lower than for those of slight and moderate overlap data. The worst classification accuracy was found for mixed data with learning rate equals to 0.1; whereas the best classification accuracy was found when the number of hidden nodes and output categories are equal. As a result, the classification accuracy increased as the number of training patterns increased. Conclusions and discussions are provided for practical guidelines.

Abstract: This paper proposes a line segment method to estimate the depth information from a pair of rectified images. The proposed method can achieve fast and high quality stereo-matching. This method first uses a simple edge detection to find out the line segments in the reference image and then calculates the color difference of each line segment from binocular images. The last step is to find out the minimum difference of each line segment as the corresponding points. From the experimental results, it is proved that the proposed method can fast and accurately generate the depth information from binocular images.