基于改进粒子群BP算法的神经网络及应用

融合粒子群算法和BP算法的优点,提出了改进粒子群BP算法,阐述了其基本思想,详细的实施步骤及对标准BP算法的改进之处。为了验证提出算法的优越性,以某钢厂引进的连铸板坯二冷动态控制系统为研究对象,设计了基于该算法的表面温度神经网络控制器。以实际生产现场的设备、工艺参数为基础进行了仿真研究,改进粒子群BP算法在收敛速度、计算精度、最优解的搜索能力、算法稳定性等方面优于标准BP算法。     

基于改进的神经网络异常声音自动识别系统研究

针对标准的BP神经网络对于声音信号识别率不高的问题,提出了一种用粒子群算法（PSO）优化BP神经网络的算法,建立了声音信号识别模型。PSO优化BP神经网络主要是用PSO来优化BP神经网络的初始权值和闽值,然后通过训练BP神经网络得到识别模型的最优解,优化后的神经网络具有误判率小、反应速度快等特点。在实验中把标准的BP神经网络和PSO优化后的BP神经网络用于八种异常声音的MFCC特征量和差分MFCC特征量识别,结果表明：在声音信号的识别系统中采用PSO优化BP神经网络的算法提高了系统的识别性能,达到了系统设计的目的。     

基于PSO-BP神经网络的地铁屏蔽门故障智能预测

为了减少地铁车站设备故障导致的人员伤亡,本文提出了粒子群算法结合BP神经网络对屏蔽门系统的故障进行预测。利用BP神经网络结构作为粒子群算法的适应度函数对BP网络的权值与阈值进行优化。在确定神经网络结构之后,该模型以权值和阈值作为粒子,利用粒子群算法的寻找全局最优的思想为BP网络寻找最优权值和阈值。减少了BP神经网络的训练结果出现较大偏差的概率。该算法可以适用于地铁站内受多种不定因素影响的设备,本文采用屏蔽门系统故障较为频繁的门锁机构来分析模型,得到的预测结果相差不到一天范围内,因此该算法具有理想的预测精度。最后利用MATLAB仿真验证该算法的可用性。     

PSO算法优化BP神经网络

本文将PSO算法与BP神经网络算法相结合,主要目的是解决BP神经网络自身存在的缺陷,利用粒子群算法来代替BP网络中梯度下降法,从而可以更好地优化BP网络各层间的连接权值,进而提高了BP神经网络的泛化能力、学习能力,并且大大提升它的收敛速度。     

粒子群算法优化BP神经网络的指纹识别技术

指纹识别技术是当今应用最广泛的生物识别技术之一。在指纹识别过程中,图像处理、特征提取、匹配等过程数据量庞大,计算比较烦琐。BP神经网络具有良好的自学习能力、强大的分类能力和容错能力,将其应用到指纹识别中是可行的。为改进BP神经网络计算速度较慢,梯度下降法不能处理一些不可微传递函数的问题,采用粒子群算法对BP算法进行优化,提高了指纹识别的速度和准确度。     

基于QPSO—BP神经网络的财务舞弊识别研究

为提高上市公司财务舞弊识别模型判断的准确度,文章以1998--2009年中国证监会网站上公开披露的财务舞弊的56家上市公司的75个舞弊年度为研究对象,并选取了与舞弊公司同行业、同年度的75个非舞弊上市公司年度作为控制样本,运用量子粒子群算法改进的BP神经网络建立财务舞弊的识别模型。研究结果表明,采用量子粒子群算法与BP神经网络结合的方法建立的财务舞弊识别模型判断的准确度较标准BP神经网络判断的准确度有了较大提高。这对于防范上市公司财务舞弊、提高监管效率、降低投资者的损失等方面具有一定的积极意义。     

数据挖掘技术在网络舆情预测中的应用

提出一种数据挖掘技术的网络舆情组合预测模型。首先利用灰色模型GM模型对网络舆情进行预测,然后采用BP神经网络对GM模型预测结果进行修正,同时采有粒子群算法优化BP神经网络参数,最后采用某网络某热门话题对模型进行仿真实验。仿真结果表明,本文模型可以准确预测网络舆情变化趋势,提高了网络舆情预测精度。     

粒子群优化BP的注塑成型工艺参数优化研究

针对注塑过程中工艺参数的优化选择问题,提出一种改进粒子群算法优化BP神经网络的成型工艺参数优化方法。将正交实验所得工艺参数作为训练样本,通过粒子群优化BP神经网络算法构建其预测模型,寻找最佳的注塑成型工艺参数。结果表明,该方法能够更快、更好地获得注塑成型中的工艺参数。     

机器人SLAM算法失真模糊性消除优化方法研究

为准确有效实现对移动机器人的同时定位和环境创建,提出将传统的粒子滤波SLAM技术和模糊神经网络算法相结合,对粒子滤波SLAM算法进行改进。提出在训练过程中设置频率计数器的方法,使测量的机器人位姿信息参数失真达到最小。实际场地实验和数据仿真结果表明通过本算法的SLAM定位信息轨迹能正确反映机器人的原始行踪轨迹,其行踪轨迹偏差的误差率仅为4.03%。与传统的粒子滤波算法相比,表现为轨迹跟踪的机器人SLAM同时定位性能比传统的粒子滤波算法提高一倍以上,仿真结果展示了新的机器人SLAM算法良好的快速收敛性能和高精度定位性能。     

基于一种改进的神经网络的RFID室内定位算法

室内定位算法精度一直都是研究的重点,本文提出了一种基于粒子群算法,减聚类算法和Kmeans算法进行结合。本文首先构建室内定位RFID模型,构建定位方程,然后采用减聚类算法来避免人为干扰,通过K-means算法来形成初始化粒子群算法,最后采用粒子群算法训练RBF神经网络的所有参数,从而得到优化的输出模型,从而确定了定位最优点。仿真实验表明本文的算法可以有效的提高定位精度,降低能量消耗,提高定位精度10%。     

Students' Reactions to the Introduction of Videoconferencing for Classroom Instruction

This study examined how students who had no prior experience with videoconferencing would react to the use of videoconferencing as an instructional medium. Students enrolled in seven different courses completed a questionnaire at the beginning of the semester and again at the end of the semester. Students at the origination and remote sites did not differ in their reactions toward videoconferencing but there was a significant difference for gender. Women reacted less favorably to videoconferencing. Compared to the beginning of the semester, students reported significantly less positive attitudes toward taking a course through videoconferencing at the end of the semester. There were no significant differences in students' attitudes toward videoconferencing across courses at the beginning of the semester but there were significant differences across the courses at the end of the semester. The results suggest the need for better preparation for both students and instructors.     

Cooler temperatures inductive to locust outbreaks

Ajoint study by Prof. ZHANG Zhibin from the CAS Institute of Zoology and his co-workers from Norway, US and Swiss have indicated that historical outbreaks of migratory locusts in China were associated with cold spells, suggesting that China＇s projected climate warming could decrease the pest＇s numbers. The study was published in Proceedings of theNational Academy of Sciences on 17 September, 2007.     

Exploring Multiplexity: Social Network Structures in a Computer-Supported Distance Learning Class

A computer-mediated group is a complex entity whose members exchange many types of information via multiple means of communication in pursuit of goals specific to their environment. Over time, they coordinate technical features of media with locally enacted use to achieve a viable working arrangement. To explore this complex interaction, a case study is presented of the social networks of interactions and media use among members of a class of computer-supported distance learners. Results show how group structures associated with project teams dominated who communicated with whom, about what, and via which media over the term, and how media came to occupy their own communication niches: Webboard for diffuse class-wide communication; Internet Relay Chat more to named others but still for general communication across the class; and e-mail primarily for intrateam communication. Face-to-face interaction, occurring only during a short on-campus session, appears to have had a catalytic effect on social and emotional exchanges. Results suggest the need to structure exchanges to balance class-wide sharing of ideas with subgroup interactions that facilitate project completion, and to provide media that support these two modes of interaction.     

The Role of Intermediating Institutions in the Diffusion of Electronic Data Interchange (EDI): How Industry Associations Intervened in Denmark, Finland, and Hong Kong

Electronic data interchange (EDI) provides means for interorganizational communication, creates network externalities, requires an advanced information technology (IT) infrastructure, and relies on standards. In the diffusion of such innovations, institutional involvement is imperative. Such institutions contain governmental agencies, national and global standardization organizations, local government, and nonprofit private organizations like industry associations. The last type of organizations we call intermediating institutions. They intermediate or coordinate ("inscribe") the activities of a group of would-be adopters. Unfortunately, little is known of how these organizations shape the EDI diffusion trajectory. In this article we examine one specific type of intermediating organizations?industry associations?and how they advanced the EDI diffusion process in the grocery sectors of Hong Kong, Denmark and Finland. We identify six institutional measures, placed into a matrix formed by the mode of involvement (influence vs. regulation) and the type of diffusion force (supply push vs. demand pull), that can be mobilized to further the EDI diffusion. Industry associations were found to be active users of all these measures to varying degrees. Their role was critical especially in knowledge building, knowledge deployment, and standard setting. Furthermore, institutional involvement varied due to policy and cultural contingencies and power dependencies.     

How Electronic Publishers are Protecting Against Piracy: Doubts About Technical Systems of Protection

The increasing prospects of digital piracy has prompted the perceived need by electronic publishers to adopt technical systems of protection, and governments to reform their copyright laws. This article is a preliminary study of the management of intellectual property by electronic publishers, defined as those involved in the production of online databases, and CD-ROMs. It focuses on three main issues: (1) how electronic publishers view the increasing threat of piracy; (2) the methods of protection employed to protect intellectual property in digital format; and (3) the importance of technological protection of intellectual property in electronic publications. The analysis is based on a sample of 23 UK electronic publishers. The interviews revealed an interesting assortment of protection methods and did not show that technological protection was a preferred way. Instead, the means of protection, in addition to copyright law, comprised niche markets, pricing, trust, bad publicity, and nontechnical and technical means.     

Rough Justice: A Statistical Assessment of ICANN's Uniform Dispute Resolution Policy

Long-standing conflict between domain name registrants and trademark holders prompted the Internet Corporation for Assigned Names and Numbers (ICANN) to create a global, mandatory arbitration procedure known as the Uniform Dispute Resolution Policy (UDRP). The UDRP has been used in 2166 cases involving 3938 domain names as of 1 November 2000. The policy gives the initiator of a complaint, generally a trademark holder, the right to choose which ICANN-accredited dispute resolution service provider (RSP) will handle the case. During the preparation of the UDRP,some feared that complainant selection would lead to "forum shopping" that might bias the results. This article performs a statistical assessment of the forum-shopping thesis and finds support for it. There are statistically significant differences in the various RSPs' propensity to take away names from defendants; there are also major differences in the number of cases brought to each RSP. RSPs who take away names have the larger share of cases. The study examines other variables that might explain differences in market share, such as price, the plaintiff's nationality, or the time taken to decide a case. It finds that nationality and time also are correlated with market share. The study concludes that shopping for a favorable outcome is an important factor in the UDRP.     

Prevalence of soft neurological signs in Chinese people

Prof. Raymond C.K. Chen, a neuropsychologist with the CAS Institute of Psychology, has made novel progress in his studies of schizophrenia. His work has been reported by a recent issue of Behavioural Neurology.     

Monoculture Rubber Plantations Have Negative Ef ects on Controlling Splash Erosion

<正>In Xishuangbanna,one of China’s most biodiverse regions,landscape has changed dramatically during the past three decades due to the conversion of tropical rainforest to rubber plantations.In steep areas,terraces are often constructed before planting rubber trees,which causes two important changes in the soil:the destabilization of soil in the bench terraces and the increased vulnerability of unvegetated riser faces to erosion.Few studies have documented the nature and intensity of erosion on bench terraces.Prof.LIU Wenjie and his colleagues from the Xishuangbanna Tropical Botanical Garden(XTBG)conducted a study in Menglun County(21°5′39″N,101°15′55″E),Xishuangbanna to evaluate the influence     

Chinese Scientists Use Earth Observation Technologies to Study and Protect Wild Camels

<正>Chinese scientists plan to apply Earth Observation technologies to protect the critically endangered wild camels(Camelus ferus).With the help of remote sensing,satellite positioning,geographical information system and wireless sensors networks,they will be able to access the distribution and population of the wild camels and protect their habitats.The project will be carried out by the International Research Center for Wild Camel Conservation,which was jointly established earlier this year by the Institute of Remote Sensing and Digital Earth(RADI)of the Chinese     

More than Skin Deep – Understand Skin Biology by Means of Dermatogenomics

<正>The Paul Gerson Unna Research Group on Dermatogenomics was founded in October 2012 at the Partner Institute for Computational Biology.The ultimate goal of the group is to understand the biology of skin and skin appendages.