Applied Mechanics and Materials Vols. 519-520

Abstract: In this paper, we thus turn to articulatory of Uyghur phonetics, the study of how phone are produced as the various organs in the mouth, throat, and nose modify the airflow from the lungs. And beside that we use a Bayesian approach based on multivariate Gaussian distribution to analyses of the acoustic articulatory of Uyghur phonemes.

Abstract: Stability, the ability to automatically extract and produce the efficient and accurate results of a defined problem without making epistemic assumptions, is discussed here as a possible memory system for understanding complex cognitive functions of the arithmetical learning. Stability is of top priority because it may typify organization of granule (knowledge-based information unit) structure. Memory efficiencies are that they depend on both linguistic factors and exposure to arithmetic training during granule formation or consolidation, supporting the idea of analog coding of numerical representations. Neuroimaging studies suggest that the parietal lobe as a potential substrate for a domain-specific representation of numeric quantities and associative memory mechanisms in stability, and results from these studies indicate that there may be the organization of number-related processes of stability in the parietal lobe. Stability seems to depend on the automatic information-processing system's response to experiential knowledge combining granularity (degree of detail or precision), maturational constraints, spatial factors (mental number line) and linguistic factors, making it an ideal candidate for understanding how these interactions play out in the cognitive arithmetic system.

Abstract: In order to improve the prediction ability of grain yield, the grain yield data of Jilin province is taken as the research object, GM(1,1) model and GM (1,N) model is established. According to the correlation analysis results, some key correlation factors of grain yield are selected into the prediction model, including the amount of chemical fertilizer, the end head of livestock, the grain sown area etc, and carries on the forecast to the grain yield of 2010-2012. The predicted results show that the average prediction error of GM(1,1) model is 6.6705% and the average prediction error of GM (1,N) model is 5.2020%. Through the comparative analysis, GM (1,N) model has higher prediction accuracy for the multiple attribute intelligent decision problem, it can be used for the prediction of grain yield in Jilin province.

Abstract: As there are many prediction problems under fuzzy environments, describing the prediction results systematically and constructing a fuzzy prediction method with good structural characteristics have attracted an extensive attention. For the prediction of investment return under fuzzy environment, we first make an analysis of general fuzzy decision-making problem, and point out its limitations. Then, we discuss the association feature between decision and membership state, and give a level effect function which can describe the recognized degree under different level cut sets. Furthermore, we establish a measure model for fuzzy optimal value based on level effect function. Finally, we apply the established model to a concrete investment example, and analyze its effectiveness in fuzzy prediction. Theoretical analysis and case study show that this method has good structural characteristics and practical significance, it can enrich the existing fuzzy prediction methods to a certain degree.

Abstract: It’s a basic work for Tibetan information processing to tag the Tibetan parts of speech,the results can be used in machine translation, speech synthesis and so on. By studying the Tibetan language grammar and the classification of Tibetan parts of speech, established the Tibetan parts of speech tagging sets, and tagged the corpus, used the CRFs to solve the problem that automatic tagging of Tibetan parts of speech, the experimental results show that in the closed test set, part-of-speech tagging accuracy is 94.2%, and in the opening set, the accuracy is 91.5%.

Abstract: The purpose of the paper is to research a fast and effective algorithm of iris localization based on Hough transform, for improving the quality of iris localization. The methods of practice include as follows. Firstly, the pupil center is estimated by using of a metric. Secondly, the binary edge image is transformed into the polar coordinates. After the rules of horizontal edge selection are used to select horizontal edge, the edge-selected image is transformed into the Cartesian coordinates. Finally, the Hough transform and the coupling relationship of the boundaries are used to determine the parameters of boundaries, and the biggest boundary parameter is selected for the estimation of the iris boundary parameter. Experiments indicate that the proposed algorithm is effective and available.

Abstract: In this paper, The notions of fuzzy abundant semigroups and fuzzy weakly abundant semigroups were introduced. On this base, some properties of fuzzy abundant semigroups and fuzzy weakly abundant semigroups were given, and some results on such semigroups were obtained.

Abstract: The sparse sets of linear equations are produced in electrical surveying, how to raise the efficiency of the solution of equations is the key to Object-probed. Glowworm swarm optimization algorithm (GSO) algorithm put forward to solve linear equations.To overcome slow convergence and lower accuracy solution, independent movement and self-adaptive step was proposed to improve the GSO (IGSO). The experimental results prove that, IGSO has a better performance than GSO.

Abstract: The framework of auto speech recognition of Lhasa dialect was established in this paper. Phoneme was chosen as the basic unit for modeling. Then, phonemes set of Lhasa dialect and their Latin transliteration were designed. There were 5568 frequently used monosyllables in the vocabulary. Hidden Markov Models of triphones were established and trained by use of HTK. Word error rate (WER) was 21.81% under the optimal situation.

Abstract: Text classification presents difficult challenges due to the high dimensionality and sparsity of text data, and to the complex semantics of the natural language. Typically, in text classification the documents are represented in the vector space using the Bag of words (BoW) technique. Despite its ease of use, BoW representation does not consider the semantic similarity between words. In this paper, we overcome the shortage of the BoW approach by applying the exponential kernel, which models semantic similarity by means of a diffusion process on a graph defined by lexicon and co-occurrence information, to enrich the BoW representation. Combined with the support vector machine (SVM), experimental evaluation on real data sets demonstrates that our approach successfully achieves improved classification accuracy with respect to the BoW approach.