A DATA MINING METHOD BASED ON CONSTRUCTIVE NEURAL NETWORKS

In this letter,Constructive Neural Networks (CNN) is used in large-scale data mining. By introducing the principle and characteristics of CNN and pointing out its deficiencies,fuzzy theory is adopted to improve the covering algorithms. The threshold of covering algorithms is redefined. "Extended area" for test samples is built. The inference of the outlier is eliminated. Furthermore,"Sphere Neighborhood (SN)" are constructed. The membership functions of test samples are given and all of the test samples are determined accordingly. The method is used to mine large wireless monitor data (about 3×107 data points),and knowledge is found effectively.

Ex vivo 3D osteocyte network construction with primary murine bone cells

Osteocytes reside as three-dimensionally（3D） networked cells in the lacunocanalicular structure of bones and regulate bone and mineral homeostasis. Despite of their important regulatory roles, in vitro studies of osteocytes have been challenging because：（1） current cell lines do not sufficiently represent the phenotypic features of mature osteocytes and（2） primary cells rapidly differentiate to osteoblasts upon isolation. In this study, we used a 3D perfusion culture approach to：（1） construct the 3D cellular network of primary murine osteocytes by biomimetic assembly with microbeads and（2） reproduce ex vivo the phenotype of primary murine osteocytes, for the first time to our best knowledge. In order to enable 3D construction with a sufficient number of viable cells, we used a proliferated osteoblastic population of healthy cells outgrown from digested bone chips. The diameter of microbeads was controlled to：（1） distribute and entrap cells within the interstitial spaces between the microbeads and（2） maintain average cell-to-cell distance to be about 19 mm. The entrapped cells formed a 3D cellular network by extending and connecting their processes through openings between the microbeads. Also, with increasing culture time, the entrapped cells exhibited the characteristic gene expressions（SOST and FGF23） and nonproliferative behavior of mature osteocytes. In contrast, 2D-cultured cells continued their osteoblastic differentiation and proliferation. This 3D biomimetic approach is expected to provide a new means of：（1） studying flow-induced shear stress on the mechanotransduction function of primary osteocytes,（2） studying physiological functions of 3D-networked osteocytes with in vitro convenience,and（3） developing clinically relevant human bone disease models.

Construction of NIM sub-network regulated by Liuwei Dihuang decoction and its characteristics

OBJECTIVE To construct the neuroendocrine immunomodulation(NIM) sub-network regulated by Liuwei Dihuang decoction(LW) and analyze its characteristics.METHODS We took the GSE57273 in GEO database and screened the differentially expressed genes(DEGs)(P<0.01) by the GEO2 R tool as gene expression signature of LW.The global PPI network was constructed in the context of whole PPI network through direct interaction algorithm and forest algorithm respectively.Then the enrichment and the topological characteristics of NIM signaling molecules were evaluated by permutation test.Finally,we abstracted the NIM sub-network by NIMNT,which combined the NIM molecular network and forest algorithm,and analyzed the topological characteristics of it by the Network Analyzer(release 2.7) plugin in Cytoscape v3.5.1.RESULTS We got 2468 DEGs in the gene expression signature of LW.After analyzing the global PPI network of LW got by two kinds of algorithms,we found that the NIM signaling molecules significantly enriched and located in important positions in the global PPI network.The NIM sub-network regulated by LW contained 1099 nodes and 1056 edges.We screened out 22 hub nodes(Degree>10).Among them,there were 19 NIM signaling molecules in which only ESR1 changed significantly and 3 non-DEGs(NFATC2,RARA,TP53).However,the down.stream of the hub nodes were significantly changes.CONCLUSION The results suggested that LW might mainly regulate the non-hub nodes to recovery of the network imbalance of the body in the state of disease.

Improving the Input of Classified Neural Networks Through Feature Construction

A general classification algorithm of neural networks is unable to obtain satisfied results because of the uncertain problems existing among the features in moot classification programs, such as interaction. With new features constructed by optimizing decision trees of examples, the input of neural networks is improved and an optimized classification algorithm based on neural networks is presented. A concept of dispersion of a classification program is also introduced too in this paper. At the end of the paper, an analysis is made with an example.

Discussion on Construction of CDMA2000 1xEV-D0 Network

The research advances on 3G CDMA standards are outlined and the network architecture and networking characteristics of CD- MA2000 1x EV-DO and CDMA2000 1x EV-DV are described.Em- phasis is put on the key technical issues that may occur when China Unicom upgrades CDMA2000 1x to CDMA2000 1x EV-DO.

Comparison of School Building Construction Costs Estimation Methods Using Regression Analysis, Neural Network, and Support Vector Machine

Accurate cost estimation at the early stage of a construction project is key factor in a project’s success. But it is difficult to quickly and accurately estimate construction costs at the planning stage, when drawings, documentation and the like are still incomplete. As such, various techniques have been applied to accurately estimate construction costs at an early stage, when project information is limited. While the various techniques have their pros and cons, there has been little effort made to determine the best technique in terms of cost estimating performance. The objective of this research is to compare the accuracy of three estimating techniques (regression analysis (RA), neural network (NN), and support vector machine techniques (SVM)) by performing estimations of construction costs. By comparing the accuracy of these techniques using historical cost data, it was found that NN model showed more accurate estimation results than the RA and SVM models. Consequently, it is determined that NN model is most suitable for estimating the cost of school building projects.

Research and Design of Network Management Construction Based on Mobile AGENT

The technical characters of mobile agent （MA） originated in the distributional artificial intelligence domain is introduced. A network management construction based on agent （NMCA） is then proposed. The NMCA structure features are elaborated in detail. A prototype design of NMCA is given by using the jKQML programming. The establishment of NMCA platform will be helpful to reduce the correspondence load of network management and improves the efficiency and the expansion ability of network management systems.

The Budget Constrained Multi-product Newsboy Problem with Reactive Production:A Problem from Entrepreneurial Network Construction

This paper develops an extended newsboy model and presents a formula- tion for this model. This new model has solved the budget contained multi-product newsboy problem with the reactive production. This model can be used to describe the status of entrepreneurial network construction. We use the Lagrange multiplier procedure to deal with our problem, but it is too complicated to get the exact solu-tion. So we introduce the homotopy method to deal with it. We give the flow chart to describe how to get the solution via the homotopy method. We also illustrate our model in both the classical procedure and the homotopy method. Comparing the two methods, we can see that the homotopy method is more exact and efficient.

Construction of GPS MonitoringNetwork,Characteristics of RecentCrustal Movement,and Relationbetween Stress Field andEarthquake Activity in the NingxiaRegion

GPS observation network is deployed in the central part of Ningxia, which is the juncture of the Alxa block, Ordos block and Qinghai-Xizang (Tibet) block. Using the data of five phases of repeated survey sine 1996, the current state of crustal movement in the central part of Ningxia is analyzed. From the result, we can know the following. (1) In the period from Dec. 1996 to May 1999, the central part of Ningxia had the phenomenon of left-lateral movement about the west margin of Ordos (measuring station P2) and the Lingwu fault on the east of the Yinchuan basin displayed the mode of left-lateral reverse strike slip movement. In that region, the direction of the principal stress field was NNE-SSW (with an azimuth of 29.8?; the central part to the south of the measuring station P2 displaced eastward; the vertical deformation was obviously greater than the horizontal deformation in order of magnitude; the Yinchuan basin and Qinghai-Xizang (Tibet) block were in a state of rising; the measuring station P1 in the hinterland of Ordos showed a trend of subsiding year by year; and there may be a hidden fault to exist between the measuring points P3 and P4. (2) About one year before the occurrence of moderately strong earthquakes in the vicinity of the measuring region, deformation anomalies and abnormal changes of principal stress direction can be observed by the GPS measuring stations in that region; before moderately strong earthquakes near the measuring region and before strong earthquakes in adjacent regions, the simulated GPS deformation vector field ofthat region can betoken the approximate position of the coming earthquake. These results can be regarded as the eigenvalues of earthquake prediction for consideration.

Self-Constructing Neural Network Modeling and Control of an AGV

Tracking precision of pre-planned trajectories is essential for an auto-guided vehicle (AGV). The purpose of this paper is to design a self-constructing wavelet neural network (SCWNN) method for dynamical modeling and control of a 2-DOF AGV. In control systems of AGVs, kinematical models have been preferred in recent research documents. However, in this paper, to enhance the trajectory tracking performance through including the AGV’s inertial effects in the control system, a learned dynamical model is replaced to the kinematical kind. As the base of a control system, the mathematical models are not preferred due to modeling uncertainties and exogenous inputs. Therefore, adaptive dynamic and control models of AGV are proposed using a four-layer SCWNN system comprising of the input, wavelet, product, and output layers. By use of the SCWNN, a robust controller against uncertainties is developed, which yields the perfect convergence of AGV to reference trajectories. Owing to the adaptive structure, the number of nodes in the layers is adjusted in online and thus the computational burden of the neural network methods is decreased. Using software simulations, the tracking performance of the proposed control system is assessed.

类风湿关节炎铁死亡特征基因CeRNA网络构建及免疫表现

背景:研究发现铁死亡相关基因在类风湿关节炎的发病机制中占据重要地位,但目前尚缺乏关于类风湿关节炎铁死亡特征基因的免疫表现及CeRNA互作网络的构建,而机器学习作为生物信息学中强大的特征基因选择算法能更精确地筛选出在类风湿关节炎发病机制中占主导地位的铁死亡特征基因。目的:利用生物信息学与机器学习方法筛选类风湿关节炎铁死亡特征基因,并分析铁死亡特征基因与免疫浸润的相关性及铁死亡特征基因CeRNA的网络构建。方法:从GEO数据库获取与类风湿关节炎相关的芯片,利用R语言提取铁死亡相关基因及其差异基因表达;使用机器学习方法对差异基因进行筛选,即运用LASSO回归与SVM-RFE方法进行特征基因筛选,对两者过滤后的基因进行再次交集,最终得到类风湿关节炎的特征基因,运用ROC曲线评估筛选后的疾病特征基因诊断疾病的准确性;利用CIBERSORT算法分析类风湿关节炎与正常滑膜组织的免疫浸润情况,并分析铁死亡特征基因与免疫细胞的相关性,最后构建类风湿关节炎铁死亡疾病特征基因的CeRNA网络并对疾病特征基因进行验证。结果与结论:①得到与类风湿关节炎相关铁死亡基因150个,其中55个上调基因,95个下调基因;②GO与KEGG富集分析分别得到18个GO显著相关条目与30个KEGG条目,主要涉及金属离子稳态、有铁离子稳态与氧化应激反应等;③机器学习分析最终获得疾病特征基因GABARAPL1、SAT1;④GSEA分析发现脂肪细胞因子信号通路、药物代谢细胞色素P450、脂肪酸代谢、PPAR信号通路、酪氨酸代谢主要集中在GABARAPL1高表达时,趋化因子信号通路、肠道免疫网络对IGA产生的影响主要集中在SAT1高表达时;⑤免疫浸润分析发现类风湿关节炎与9种免疫细胞与正常组织存在明显差异,其中浆细胞、T细胞CD8、T细胞滤泡辅助器在疾病组中为高表达状态,其余为低表达。单基因与免疫细胞相关性分析发现GABARAPL1在树突状静息细胞、激活的NK细胞、巨噬细胞M1等为正相关,其中与树突状静息细胞的相关性最为显著,SAT1与T细胞CD4与γδT细胞为正相关,与NK静息细胞为负相关;⑥GSVA分析发现抗坏血酸和醛酸代谢在SAT1表现为上调的基因高表达水平时候为上调,而B细胞受体信号通路、TOLL样受体信号通路、T细胞受体信号通路、自然杀伤细胞介导的细胞毒性等表现为下调,PPAR信号通路、烟酸盐和烟酰胺的代谢、色氨酸代谢、脂肪酸代谢、类固醇生物合成等在GABARAPL1表现为下调趋势;⑦60种长链非编码RNA可能在导致类风湿关节炎发展过程中发挥关键作用。提示:类风湿关节炎的发生与类风湿关节炎铁死亡疾病特征基因的异常表达显著相关,特征基因通过影响相关信号通路诱导疾病的发生发展,并通过对类风湿关节炎相关长链非编码RNA介导的ceRNA网络进行分析,识别出潜在的治疗靶点及信号通路,为进一步阐明其发病机制,并为后续的实验研究提供参考依据。

新型电力系统主动构网机理与技术路径

随着新一代“双高”电力系统的发展,新能源并网变换系统逐渐由随网型向构网型演变,当新能源机组集群运行时,并网情况将更加复杂。变换器的构网能力需亟待研究,同时也是电力能源转型需承担的重要使命。该文立足并网系统内外形态的全局性和传统并网变换系统性能分析的局限性,对并网变换器的主动构网支撑机理进行再揭示。首先,对虚拟同步技术与构网机理进行简要评述;其次,就集群并网系统的建模问题、多机集群并网运行时的耦合交互问题、支撑构网能力的储能配置和运行边界等3个关键技术问题,进行深入研究;最后,针对上述3个关键技术问题,进行技术路径探索和研判。该文的思路为形成适用于未来新能源集群并网运行的系统化理论和实用化方法奠定一定基础。

基于动态贝叶斯网络的装配式建筑吊装施工安全风险分析

针对装配式建筑吊装施工期间风险影响因素多、不确定性高及风险状态随时间动态变化等问题,提出基于动态贝叶斯网络的装配式建筑吊装施工安全风险动态分析方法。首先,根据装配式建筑吊装施工特点和事故致因理论,从“人、物、环、管”4个方面建立风险指标体系。其次,引入时间维度建立动态贝叶斯网络模型,利用模糊理论和专家打分法量化网络节点的概率,并结合Leaky Noisy-or Gate扩展模型修正条件概率。最后,利用动态贝叶斯网络的双向推理功能对装配式建筑吊装施工安全风险进行动态风险分析。由实例分析得到某装配式建筑吊装施工安全风险的时序变化曲线,通过反向推理得到导致吊装事故发生的关键风险因素。研究成果可为分析和有效控制装配式建筑吊装施工安全风险提供新思路。

网络刑法的发展回溯与理论供源

传统刑法体系遭遇网络犯罪形态后频现应答疲态,预防性立法由此显身,开启由网络犯罪策动刑法转型的号角,但网络刑法的理论根基尚不稳。网络刑法的演进是网络社会发展与网络犯罪迸发共同作用下真实的渐显;网络空间社会思维是网络刑法的建构基石,衍生丰富的刑法学意义。网络安全秩序价值的相对优位与网络自由的并重、网络安全法益的统领地位与刑法任务变更、预防性理念的内在觉醒与审慎运行,是网络刑法的根基之本,框定网络刑法的发展脉络。刑法知识迁移是渐进的,网络犯罪、网络归责、网络制裁堪当生成网络刑法知识范畴的参照性“母版”,未来可创设网络刑法典以实现知识结构的创造性立法固化。

社工机构介入儿童友好型社区营造的行动逻辑与可持续发展策略——以安徽省X项目为例

儿童友好型社区营造需要多主体协同参与,包括由民政部门、基层社区、社区居民、高校等主体组成的人类行动者以及社区文化、公共空间、课程体系等非人类行动者,社工机构如何激活这一行动者网络依然有待讨论。本文以安徽省X项目为例,借鉴行动者网络理论的框架,从转译的不同阶段分析社工机构的行动逻辑:在问题呈现阶段确立共同行动目标与强制通行点;在利益赋予阶段协调各类行动者的关键利益;在征召阶段推动异质行动者的融合加入;在异议动员阶段构建互利共赢的利益共同体。为了保障儿童友好型社区营造的可持续性,社工机构可采取的策略包括:通过调研评估挖掘社区特色的内源力、通过跨界融合吸纳异质行动者的加入、通过平台搭建畅通主体间的沟通渠道、通过充权赋能保障规划运营的自主化。

陆域大地电磁基础网规划与展望

阵列式大地电磁测深是区域地球物理调查的重点发展方向之一,对标国外先进水平和国家社会的基础性需求,中国阵列式大地电磁测深工作程度依然较低,缺乏整体规划,开展全国性观测计划亟需提上日程。系统总结了当前区域地球物理学的发展趋势及国内外大地电磁网建设现状,提出了合作共建陆域多层级大地电磁基础网和大数据中心的顶层设计和规划布局,并对中国大地电磁基础网未来的重点发展方向进行了展望和思考。相关工作的顺利开展将为中国的区域地球物理调查拓展一项新属性参数,推动构建物性地球、结构地球和地球物理场四维模型,进而变革区域地球物理研究范式和工作模式,支撑中国深地探测走向世界前列。