Applied Mechanics and Materials Vols. 263-266

Abstract: Due to the complex nonlinear characteristics between erosive rainfall and corresponding sediment volume, radial basis function (RBF) neural network is adopted to predict siltation in matlab2010 environment, and the results were compared with that one from BP neural network. In the course, the 3 major indicators of a rainfall such as single rainfall erosivity (R), maximum rainfall intensity in thirty minutes (I30) , rainfall quantity(P) are as input vectors, with the actual sediment deposition as a target vector. The results show that: RBF neural network is better than BP neural network in forecasting accuracy, computation speed, fitting accuracy.

Abstract: Based on the analysis of the flaw mechanism of tradition genetic algorithm and critically inheriting the immunity theory foundation, the paper proposes a new optimized algorithm - Subregion Artificial immunity Optimized Algorithm Based on Parallel, which can speed up the partially convergence and at the same time maintain the global convergence through the simulation of actual immunity behavior of organism. The basic thought of this algorithm is firstly to divide a complex question into several simple questions, secondly to carry out parallel “immunity” computation according to those simple questions, thirdly to pour the results (the antibodies) into the global immunity algorithm, which not only simplifies the complex question, but also raises the efficiency of the algorithm.

Abstract: When a person watches different marrow-cell images he or she can identify every type of cells easily. In this process, human’s visual system has ability to adapt the different shades of the color marrow cells images. We propose a segmentation method for marrow-cell images based on fuzzy c-means clustering (FCM). Firstly, the count of cluster is calculated out using the shades of the R-matrix of a RGB formatted marrow cells image. Secondly, the fuzzy c-means clustering method is done on the R-matrix. Finally, the pixel of G-matrix and B-matrix are divided into some clusters by “one to one correspondence” of the position of pixels that belong to R-matrix, G-matrix or B-matrix. This paper’s contribution could be summarized into three points: 1) a frame work of the fuzzy c-means clustering for marrow-cell images segmentation is proposed. 2) Using FCM and the R- matrix component of a RGB formatted marrow-cell images to generate the count of clustering. 3) This method could adaption different shades of different marrow-cell images.

Abstract: CDIO is a kind of innovation mode of international engineering education, and CDIO capability evaluation is the main content implemented by this mode. In this thesis, the model of CDIO capability evaluation expert system is designed, and Fuzzy Petri Net technique is applied to conduct knowledge presentation and inference of inference engine. Besides, the construction of knowledge database and the algorithm and realization of inference mechanism are elaborated. CDIO capability level of the students can be evaluated intelligently by applying evaluation expert system, which can also feed back the existing key questions. In this way, the development and evaluation of CDIO capability are constituted into a loop, which improves the quality of talent training effectively.

Abstract: Regression testing is an important activity to ensure the quality of software. In order to improve the efficiency of regression testing, in this paper, the author proposes to reorder test suite based on ant colony algorithm in regression testing, and compare the result with other common sort results. Through experiment, it shows that the method can get the optimal sequence of test cases under the time limit and it has been proven a superior method in both effectiveness and efficiency for test cases prioritization.

Abstract: A probabilistic neural network (PNN) speech recognition model based on the partition clustering algorithm is proposed in this paper. The most important advantage of PNN is that training is easy and instantaneous. Therefore, PNN is capable of dealing with real time speech recognition. Besides, in order to increase the performance of PNN, the selection of data set is one of the most important issues. In this paper, using the partition clustering algorithm to select data is proposed. The proposed model is tested on two data sets from the field of spoken Arabic numbers, with promising results. The performance of the proposed model is compared to single back propagation neural network and integrated back propagation neural network. The final comparison result shows that the proposed model performs better than the other two neural networks, and has an accuracy rate of 92.41%.

Abstract: The first step of the association rule mining algorithm Apriori generate a lot of candidate item sets which are not frequent item sets, and all of these item sets cost a lot of system spending. To solve this problem，this paper presents an improved algorithm based on Apriori algorithm to improve the Apriori pruning step. Using this method, the large number of useless candidate item sets can be reduced effectively and it can also reduce the times of judge whether the item sets are frequent item sets. Experimental results show that the improved algorithm has better efficiency than classic Apriori algorithm.

Abstract: In this paper, the author did the sentiment classification experiment with Chinese hotel reviews from internet based on semantic lexicon and Naive Bayesian. Many experiments have shown the method in order to improve the accuracy of sentiment classification, and time shortened of text processing, the method can also be used in fast sentiment classification for mass data of Chinese texts.

Abstract: Sensory analysis has an important impact on food production since its results can help the understanding of consumers’ perceptions about the products. Thus, many methods have been proposed and applied to quantify sensory attributes of food products. In this paper we proposed a methodology, using Kohonen's Self-Organizing Maps and K-means algorithm, to classify food samples through the responses, provided by human evaluators, for their attributes such as aroma, flavor, appearance and texture. Conducted experiments in sensory analysis to determine the acceptance of new gelatins produced from chicken feet and new wines produced from spares of Açaí and Cajá confirm that proposed methodology is suitable for the investigated purpose.

Abstract: Conventional color-printing systems often use three or more inks, such as CMY, CMYK or CMYKLcLm. When the inks hues exceeds three, such as CMYKRGB, CMYKOrG and so on, there is the usual color-management one-to-many mapping problem. An algorithm was developed for multi-inks printing in which the one-to-many mapping problem was overcome by dividing the standard color space into several sub-spaces, founding the relationships between the sub-spaces and the printing color-separations and building the lookup table. The algorithm was tested using a digital inkjet printer-Mutoh8000 of CMYKOrGB. Mutoh8000 prints separated color blocks using this algorithm were compared with a generic ICC profile for CMYKLcLm prints. The CMYK inks were common to both prints. The algorithm has been proved effective and improved color printing quality significantly.