A neural network model approach to athlete selection

In order to determine the usefulness of neural models in optimisation of recruitment processes, statistical analyses were carried out on measured results of javelin throwers using a full take off. A group of 140 Polish junior javelin throwers took part in the research. In order to choose the optimum combination of model parameters the Hellwig method was used. Linear and multilayer perceptron neural models were constructed and used to calculate combinations of variables. Statistical analysis of the results showed that the linear model was not able to describe precisely the relationship between the dependent variable and independent variables for the investigated group of young javelin throwers. For the investigated group, the perceptron network with a 4-3-2-1 structure gave the best predictive relationship for sports results of the javelin throwers.     

基于文本蕴涵的受限领域自动问答方法研究

本文提出了一种用于受限领域自动问答的新方法.与传统的自动问答方法不同,该方法不对用户问题进行语言分析以生成结构化的查询,而是利用文本蕴涵技术从自动生成的假设问题库中寻找用户问题所蕴涵的假设问题,然后利用该假设问题所附的答案指南获取用户问题的答案.该方法的优点之一是不需要复杂的语言处理就可以回答复杂的问题,而且正确率较高;优点之二是可以利用本体方便地生成不同语种的假设问题,实现跨语言问答.本文基于英文提问对该方法进行了测评,测评结果表明几乎所有的用户问题都可以基于生成的假设问题来直接或间接地回答,回答的正确率达到65.6%.     

Comprehension monitoring and reading comprehension in bilingual students

This study explored comprehension monitoring, use of reading strategies and reading comprehension of bilingual students at different levels of perceived proficiency in Italian. The participants were bilingual fifth to eighth‐grade elementary school students from four Italian schools in Rijeka, Croatia. Students' reading comprehension was assessed. Their comprehension monitoring skill was measured on the Metacomprehension test and through use of a cloze task. The Strategic Reading Questionnaire (SRQ) was used as a self‐report measure of strategic reading. A questionnaire investigating Italian language use and perceived proficiency in the Italian language was also administered. Perceived proficiency in Italian was not clearly determined by early or late preschool age of second language acquisition. Bilingual students with high perceived proficiency in Italian (high PP group) had better meta‐cognitive reading skills than those with low perceived proficiency in Italian (low PP group). Comprehension monitoring was the most important predictor of reading comprehension in all students.     

On relative degree,chattering and fractal nature of parasitic dynamics in sliding mode control

With respect to relative degree and chattering in sliding mode (SM) control systems, the notion of fractal dynamics is introduced, and a conjecture is formulated that the character of parasitic dynamics of real control systems is fractal. A model of fractal dynamics is proposed. The characteristics of fractal dynamics are studied in the frequency and time domains. It is shown that with fractal parasitic dynamics SM control systems will always feature chattering and non-ideal closed-loop performance. An example of analysis is provided.     

Finding exact periodic solutions in Lure systems through periodic signal mapping

A solution of the periodic problem in a nonlinear system comprising a single-valued symmetric nonlinearity (Lure system) and linear dynamics is presented. The solution is designed as an iterative application of Periodic Signal Mapping to refine the approximate solution obtained through the describing function method. The algorithm is based upon the transformation of the original nonlinear system into an equivalent nonlinear system for which the filtering hypothesis is satisfied exactly, and for that reason the latter being suitable for application of the developed algorithm. The solution sought for is a fixed point of the periodic signal mapping. Conditions of asymptotic convergence of the proposed algorithm are given. The proposed approach is illustrated by two examples of analysis of periodic motions in a relay feedback system and chattering in a terminal sliding mode.     

基于GA-BP神经网络的扬声器音盆性能预测技术

在电声零器件制造行业中,扬声器音盆的性能好坏直接关系到扬声器产品的质量.本文提出用遗传算法(GA)优化BP神经网络模型对扬声器音盆性能进行预测.在分析了BP神经网络原理的基础上,主要阐述了如何应用遗传算法优化BP神经网络,以改进BP神经网络收敛速度慢、易陷入局部极小值的缺点.针对音盆产品特点,建立音盆性能的GA-BP神...     

Protein identification and localization using mass spectrometry and staining tests in cross-sections of polychrome samples

The identification and localization of the proteinaceous binders are essential issues in studies of painting materials and techniques, for further proposing valid restoration and conservation treatments of the painted or polychrome works of art. The challenge for analytical chemists and conservation scientists is the availability of methods able to simultaneously identify and map the presence of the binders in the multilayered structure of a sample and the possibility to use a very low amount of sample from the studied art object (considering also the criteria of minimum sampling). These methods should be fast, reproducible in different artefacts and in case of mixture of protein-based binders with other non-proteinaceous constituents (oils, resins, waxes, gums etc.) and also economical (both in terms of materials and time consume). In this context, the present paper proposes an innovative protocol of investigation using two complementary techniques – Matrix-Assisted Laser Desorption/Ionisation – Time of Flight Mass Spectrometry (MALDI-TOF MS) and staining tests (one visible and one fluorescent stain) assisted by Optical Microscopy (OM) on cross-section of samples – for the simultaneous identification and mapping of protein – and oil-based binders in paint materials. The novelty is based on the use of MALDI-TOF MS on cross-sections of paints together with a fluorescent stain for protein identification and mapping (mainly used in the area of proteomics) complementing the use of a traditional visible stain for oil-based material identification. The protocol was successfully applied on several samples taken from a Czech medieval polychrome sculpture, entitled “The Mourning of Jesus Christ” (16th century) belonging to the Moravian Gallery (Brno).     

面向关联数据的高校机构知识库构建方法研究

[目的/意义]针对当前我国高校机构知识库研究与应用离真正的开放互连、知识集成、知识检索还有较大差距的问题,提出一种面向关联数据的机构知识库构建方法,该方法覆盖机构知识库构建的整个过程,并支持机构知识的资源整合、语义检索、知识推理和关联数据发布,为各高校机构知识库建设提供参考。[方法/过程]调研和剖析国内外典型的机构知识库,将机构知识库构建方法分为5个步骤,包括:①评价和选择有价值的信息资源;②建立机构知识库概念模型;③进行语义描述和关联设置;④发现并创建关联化数据环境下的语义推理规则;⑤设计系统架构,并通过西南科技大学机构知识库构建实例来验证该方法的有效性。[结果/结论]实例验证表明,面向关联数据的机构知识库构建方法具有合理性和有效性,起到了降低机构知识库开发难度和缩短开发周期的作用,对于其他机构知识库的建设具有参考价值。     

生态社会的呼吁

我们需要创造一个走向生态纪元的社会!这种创造产生手布莱恩·斯维姆和托马斯·柏励所撰写的宇宙故事：“地球共同体的未来,在某种意义上,依赖于人类所做出的决定。人类已经把自己纳入地球过程的遗传密码之中,而且其程度是如此之深。推崇技术纪元的人,把地球看作资源,未来应该增加对地球的利用,所有一切都应该以人类的利益为中心：推崇生态纪元的人,提倡一种新型的人类与地球的关系,地球共同体的整体安宁是其根本的关注。生态纪元的未来可以解决这两者之间所产生的紧张状态。”