Applied Mechanics and Materials Vols. 651-653

Abstract: This paper describes rough neural network which consists of a combination of rough neurons and conventional neurons. Rough neurons use pairs of upper and lower bounds as values for input and output. In some practical situations, it is preferable to develop prediction models that use ranges as values for input and/or output variables. Integrating rough set theory with neural network theory, a novel information fusion method based on rough neural network is proposed to fuse the different-source images in agricultural robot. It is used to fuse infrared and visible images in order to take full advantage of the complementary information between infrared and visible images. Experimental results show that the fusion effect and speed are both better than standard wavelet transform and the conventional neural network.

Abstract: Product quantization (PQ) is an efficient and effective vector quantization approach to fast approximate nearest neighbor (ANN) search especially for high-dimensional data. The basic idea of PQ is to decompose the original data space into the Cartesian product of some low-dimensional subspaces and then every subspace is quantized separately with the same number of codewords. However, the performance of PQ depends largely on the distribution of the original data. If the distributions of every subspace have larger difference, PQ will achieve bad results as shown in our experiments. In this paper, we propose a uniform variance product quantization (UVPQ) scheme to project the data by a uniform variance projection before decompose it, which can minimize the subspace distribution difference of the whole space. UVPQ can guarantee good results however the data rotate. Extensive experiments have verified the superiority of UVPQ over PQ for ANN search.

Abstract: It is very important in developing effective and efficient traffic engineering schemes for identifying elephant flows. In this paper, according to the base idea of Sample and Hold, a new algorithm is proposed to realize elephant flows identification. Compare to the original sample and hold algorithm, our algorithm save storage space and reduce false positive by handling large flows firstly, using Scalable Bloom Filter to identify sampled flows and using Multilayer Compressed Counting Bloom Filter to count flows. The theoretical analysis and the simulation result indicate that under the condition of existing some tolerable measurement error about the length of flows, our algorithm can identify elephant flows accurately, which reduce the storage space and improve the processing speed efficiently.

Abstract: In the process of modeling of 3D dynamic images, influenced by the large volume of the objects, it is difficult for the traditional method to build a model really show the effects of 3D dynamic images. To solve this problem, the paper proposes a new fast modeling method for large-scale 3D dynamic images through the usage of OpenGL. It promotes the design result from 2D space to 3D space, thus making the images more lifelike. The experiment shows that the 3D dynamic images modeling algorithm can rapidly build the 3D model with excellent effect saving time and improving work efficiency.

Abstract: Unreasonable parameters may lead to a phenomenon of the numerical instability in evolutionary structural optimization method (ESO). In this paper the improved SIMP-based ESO method for the structural compliance sensitivity is presented to solve the problem of checkerboard pattern. The method depends on the sensitivity analysis results that indicate the contribution of each unit for the whole structural performance to delete and to add elements. At the same time, the method in combination with a sensitivity redistribution technology of controlling checkerboard pattern is used to realize that each element’s contribution or impact factor of the whole structural performance has a smooth transition. The instance shows that the method is reasonable and different parameters will affect the optimized results. The optimal values of parameters can be seen obviously finally.

Abstract: With the aim to meet the requirements of multi-directional choice, the paper raise a new approach to the invariant feature extraction of handwritten Chinese characters, with ridgelet transform as its foundation. First of all, the original images will be rotated to the Radon circular shift by means of Radon transform. On the basis of the characteristic that Fourier transform is row shift invariant, then, the one-dimensional Fourier transform will be adopted in the Radon domain to gain the conclusion that magnitude matrixes bear the rotation-invariance as a typical feature, which is pretty beneficial to the invariant feature extraction of rotation. When such is done, one-dimensional wavelet transform will be carried out in the direction of rows, thus achieving perfect choice of frequency, which makes it possible to extract the features of sub-line in the appropriate frequencies. Finally, the average values, standard deviations and the energy values will form the feature vector which is extracted from the ridgelet sub-bands. The approaches mentioned in the paper could satisfy the requirements from the form automatic processing on the recognition of handwritten Chinese characters.

Abstract: Well-developed repent stems grown on Pooideae, when it is growing, the new unit will also grow on their nodes, the two growing process is parallel. To research the parallel growth process in the coordinate system, we collected much growth data, observed and studied them, then put forward the parallel growth model based on state-space, the model can reflect parallel growing of repent stem continually and dynamically, and it is also easily to understand. It provide a new research method for visual plant modeling.

Abstract: This paper presents an algorithm based on related selection for reversible logic synthesis, and the algorithm is optimized. The algorithm realizes the synthesis of the whole 3-varibles functions and some part of 4-varibles functions. The algorithm for the space complexity is O(n*2ⁿ). Compared with other algorithm for reversible logic synthesis at home and abroad, this algorithm has a less gate number in the synthesis of the whole 3-varibles functions and some examples in benchmark.

Abstract: In recent years, the prevailing topic crawler algorithms are concentrated on the contents of topical words. These existing approaches neglect the sematic relationship among textual concepts, which lead to low correlation between crawled webpages. To address the issue, this paper presents a deep analysis of Shark Search algorithm, and makes an optimization in terms of incorporating the characteristics associated with semi-structured webpages. Furthermore, we enhance the performance of vector space model utilized in Shark Search algorithm by virtue of domain ontology, and propose a standardized method based on the vector space of ontology model to improve the evaluation metric of TF-IDF. The experimental results demonstrate the effectiveness of our algorithm that outperforms the state-of-the-art significantly in precision and recall.

Abstract: Term frequency/inverse document frequency (TF-IDF) is widely used in text classification at present, which is borrowed from Information Retrieval. Based on this conventional classical TF-IDF formula, we present a new TF-IDF weight schemes named CTF-IDF. The experiment shows that the improved method is feasible and effective. Furthermore, from the subsequent evaluations using 10-fold cross-validation, we can see the CTF-IDF greatly improves the accuracy of text classification.