Improving the generalization performance of RBF neural networks using a linear regression technique

In this paper we present a method for improving the generalization performance of a radial basis function (RBF) neural network. The method uses a statistical linear regression technique which is based on the orthogonal least squares (OLS) algorithm. We first discuss a modified way to determine the center and width of the hidden layer neurons. Then, substituting a QR algorithm for the traditional Gram–Schmidt algorithm, we find the connected weight of the hidden layer neurons. Cross-validation is utilized to determine the stop training criterion. The generalization performance of the network is further improved using a bootstrap technique. Finally, the solution method is used to solve a simulation and a real problem. The results demonstrate the improved generalization performance of our algorithm over the existing methods.     

字符分布特征对带有通配符串匹配问题的影响

近年来,字符串匹配问题被不断扩展。其中,具有代表性的是在模式中引入可变长度的通配符,本文称之为PMWL问题。针对此问题,已有工作分析了在不同的模式特征下,匹配数Ω随文本长度增加呈指数级增长。本文同时考虑文本分布特征和模式特征,建立了期望模型E(Ω)＝n*D*π(P),其中n为文本长度,D为模式中各通配符跨度的乘积,π(P)为基于字符分布的模式出现概率。实验部分,在人工随机数据和DNA真实数据上验证了E(Ω)的准确性,得到预测误差率分别为1.8%~3.2%和4.7%~7.8%;在不同字符分布中,分析了模式模长和通配符跨度对匹配数Ω的影响。E(Ω)模型揭示了Ω的增长趋势不一定呈指数级,而取决于π(P)和D的共同影响。此外,E(Ω)模型能够在线性时间内得到近似完备解。     

基于非合作博弈的OFDMA-WLAN系统资源分配算法研究

为满足无线局域网中基于正交频分多址技术(OFDMA-WLAN)的多用户通信需求,提出了一种基于非合作博弈的资源分配策略,利用博弈论这一有效工具对无线网络服务质量的资源分配问题进行研究,同时将用户对信道质量的不同需求转化为多用户对不同信道资源分配的非合作博弈竞争问题。此外,还将纳什均衡问题(NEP)细分为数个变分不等式子问题,通过凸优化函数对这些子问题分别求解。数值分析结果表明,所提算法在资源分配的公平性和数据传输率方面达到了更好的折中。     

Digital filters for clustered-OFDM-based PLC systems: Design and implementation

This work aims to investigate the design and implementation of digital filters for separating clusters in a clustered-orthogonal frequency-division multiplexing scheme for power line communication systems. In this regard, we formulate the mathematical problem and present a criterion which is capable of searching for digital filters that maximizes data rate. In the following, several finite impulse response and infinite impulse response digital filters are analyzed in order to validate the proposed criterion and to verify the best one which fulfills the given constraints. Furthermore, regarding only the best digital filters obtained, a finite precision and complexity analyses are carried out by using a field-programmable gate array device. Based on numerical results, we show that finite impulse response equiripple minimum phase and infinite impulse response Chebyshev type II digital filters yield the best data rates among the chosen digital filters. Moreover, we point out that equiripple minimum phase digital filters consume more hardware resources than Chebyshev type II digital filters, although the first one deal better with the field-programmable gate array constraints when more bits are used to implement the digital filter. Due to that, finite impulse response digital filters are more indicated to be implemented in a field-programmable gate array device.     

Design optimization of resource combination for collaborative logistics network under uncertainty

Collaborative logistics networks (CLNs) are considered to be an effective organizational form for business cooperation that provides high stability and low cost. One common key issue regarding CLN resource combination is the network design optimization problem under discrete uncertainty (DU-CLNDOP). Operational environment changes and information uncertainty in network designs, due to partner selection, resource constrains and network robustness, must be effectively controlled from the system perspective. Therefore, a general two-stage quantitative framework that enables decision makers to select the optimal network design scheme for CLNs under uncertainty is proposed in this paper. Phase 1 calculates the simulation result of each hypothetical scenario of CLN resource combination using the expected value model with robust constraints. Phase 2 selects the optimal network design scheme for DU-CLNDOP using the orthogonal experiment design method. The validity of the model and method are verified via an illustrative example.     

Some sets of orthogonal polynomial kernel functions

Kernel methods provide high performance in a variety of machine learning tasks. However, the success of kernel methods is heavily dependent on the selection of the right kernel function and proper setting of its parameters. Several sets of kernel functions based on orthogonal polynomials have been proposed recently. Besides their good performance in the error rate, these kernel functions have only one parameter chosen from a small set of integers, and it facilitates kernel selection greatly. Two sets of orthogonal polynomial kernel functions, namely the triangularly modified Chebyshev kernels and the triangularly modified Legendre kernels, are proposed in this study. Furthermore, we compare the construction methods of some orthogonal polynomial kernels and highlight the similarities and differences among them. Experiments on 32 data sets are performed for better illustration and comparison of these kernel functions in classification and regression scenarios. In general, there is difference among these orthogonal polynomial kernels in terms of accuracy, and most orthogonal polynomial kernels can match the commonly used kernels, such as the polynomial kernel, the Gaussian kernel and the wavelet kernel. Compared with these universal kernels, the orthogonal polynomial kernels each have a unique easily optimized parameter, and they store statistically significantly less support vectors in support vector classification. New presented kernels can obtain better generalization performance both for classification tasks and regression tasks.     

An approximation algorithm dependent on edge-coloring number for minimum maximal matching problem

We consider the minimum maximal matching problem, which is NP-hard (Yannakakis and Gavril (1980) [18]). Given an unweighted simple graph G=(V,E), the problem seeks to find a maximal matching of minimum cardinality. It was unknown whether there exists a non-trivial approximation algorithm whose approximation ratio is less than 2 for any simple graph. Recently, Z. Gotthilf et al. (2008) [5] presented a -approximation algorithm, where c is an arbitrary constant.In this paper, we present a -approximation algorithm based on an LP relaxation, where χ^′(G) is the edge-coloring number of G. Our algorithm is the first non-trivial approximation algorithm whose approximation ratio is independent of |V|. Moreover, it is known that the minimum maximal matching problem is equivalent to the edge dominating set problem. Therefore, the edge dominating set problem is also -approximable. From edge-coloring theory, the approximation ratio of our algorithm is , where Δ(G) represents the maximum degree of G. In our algorithm, an LP formulation for the edge dominating set problem is used. Fujito and Nagamochi (2002) [4] showed the integrality gap of the LP formulation for bipartite graphs is at least . Moreover, χ^′(G) is Δ(G) for bipartite graphs. Thus, as far as an approximation algorithm for the minimum maximal matching problem uses the LP formulation, we believe our result is the best possible.     

Strong matching preclusion

The matching preclusion problem, introduced by Brigham et al. [R.C. Brigham, F. Harary, E.C. Violin, and J. Yellen, Perfect-matching preclusion, Congressus Numerantium 174 (2005) 185-192], studies how to effectively make a graph have neither perfect matchings nor almost perfect matchings by deleting as small a number of edges as possible. Extending this concept, we consider a more general matching preclusion problem, called the strong matching preclusion, in which deletion of vertices is additionally permitted. We establish the strong matching preclusion number and all possible minimum strong matching preclusion sets for various classes of graphs.     

Environmental effects on the coevolution of pursuit and evasion strategies

The game of tag is frequently used in the study of pursuit and evasion strategies that are discovered through competitive coevolution. The aim of coevolution is to create an arms race where opposing populations cyclically evolve in incremental improvements, driving the system towards better strategies. A coevolutionary simulation of the game of tag involving two populations of agents; pursuers and evaders, is developed to investigate the effects of a boundary and two obstacles. The evolution of strategies through Chemical Genetic Programming optimizes the mapping of genotypic strings to phenotypic trees. Four experiments were conducted, distinguished by speed differentials and environmental conditions. Designing experiments to evaluate the efficacy of emergent strategies often reveal necessary steps needed for coevolutionary progress. The experiments that excluded obstacles and boundaries provided design pointers to ensure coevolutionary progress as well as a deeper understanding of strategies that emerged when obstacles and boundaries were added. In the latter, we found that an awareness of the environment and the pursuer was not critical in an evader’s strategy to survive, instead heading to the edge of the boundary or behind an obstacle in a bid to ‘throw-off or hide from the pursuer’ or simply turn in circles was often sufficient, thereby revealing possible suboptimal strategies that were environment specific. We also observed that a condition for coevolutionary progress was that the problem complexity must be surmountable by at least one population; that is, some pursuer must be able to tag an opponent. Due to the use of amino-acid building blocks in our Chemical Genetic Program, our simulations were able to achieve significant complexity in a short period of time.

基于多模板匹配和可信度分析的中文文档图像关键词过滤方法*

针对互联网中文文档图像非法信息过滤提出了一种多模板匹配结合可信度分析的方法。该方法克服了传统OCR速度慢的缺点,同时改善了基于图像特征匹配方法对字体和噪音敏感的特性。通过改变关键词搜索方式有效地减小了计算量,提高了识别速度。实验结果表明了该方法的有效性。