基于BP神经网络的文档特征表示研究

本文根据BP神经网络的函数逼近功能 ,针对文档特征项在文档中的权重 ,提出了一种基于BP神经网络的网络计算模型。实验表明 ,在面向相似主题的文档集中 ,这种方法比当今最常用的向量空间模型计算的文档特征项的权重更精确     

Participatory action research and impact: an ecological ripples perspective

Assessment of the impact of varied forms of participatory research is enhanced by specific attention to the ripples, such work initiates in the social context. Ripples are defined as consequences either unintended or unanticipated by those carrying out the participatory research. Following a brief documentation of the range of reported ripples in reports describing participatory research, an ecological perspective drawing on the interdependence principle of ecological systems theory is presented as a framework for assessing impact by attending to ripples in the local context. Community asset mapping and social network analysis are described as examples of ways of describing and documenting environmental ripples, and strategies for both anticipating and reviewing ripples post hoc are discussed. The importance of developing cognitive maps of local ecology is stressed as critical to the assessment of the multiple potential impacts of participatory research.     

The UK National Arts Education Archive: Ideas and Imaginings

The National Arts Education Archive (NAEA) is housed and maintained by the Yorkshire Sculpture Park (YSP), and managed by YSP coordinators and educators with a well‐established volunteer programme. This year, 2017, as part of the celebrations of the YSP's 40th anniversary, the Archive will hold its own exhibition entitled Treasures Revealed: a collection of items selected by people who have been involved in the Archive, whether as donors, volunteers, researchers, artists, trustees or steering group members. In parallel with the exhibition, this article aims to give voice to a selection of individuals and groups associated with the Archive, discussing their interests and experiences of it, and their thoughts on its value and importance as a repository of arts education materials, ideals and practices. Our primary motivations were to consider these different voices in relation to the purpose, direction and relevance of the NAEA today. These exchanges raise fundamental questions and debates about what art education is and what it might become, and how these historical collections, and creative engagements with it, might help to shape our contemporary thinking.     

基于TM遥感图像的流域地貌研究

利用不同地物在同一波段或相同地物在不同波段具有不同色调的变化规律，以奉化江流域为例，探讨了遥感技术在流域地貌研究中的应用。根据遥感图像成像规律以及不同地质地貌景观类型在TM图像上所显示的不同解译标志和展布特征，对研究区水系的发育、地质构造与地表物质对流域地貌形成的作用和影响等作了一定的分析与探索，获得了较为满意的结果。     

管理干部综合素质模糊评价体系的神经网络模型

本文根据人工神经网络的理论，利用高泛化性能的BP神经网络建立了企业管理人员综合素质的模糊评价模型，通过仿真和实例表明了模型的有效性，给管理干部综合素质的测评提供了一种简便、有效的方法。     

脑科学与类脑研究概述

脑科学和类脑智能技术是两个重要的前沿科技领域;二者相互借鉴、相互融合的发展是近年来国际科学界涌现的新趋势。脑科学研究对大脑认知神经原理的认识,提升了人类对自身的理解和脑重大疾病的诊治水平,也为发展类脑计算系统和器件、突破传统计算机架构的束缚提供了重要的依据。文章概括性地总结了国际脑科学和智能技术领域的总体进展,未来15年的发展趋势和有望解决的重大科学问题,以及我国在这两领域发展的特色和优势。     

核磁共振永磁体无源匀场的神经网络方法

针对核磁共振永磁体的磁场传统匀场方法难度大，经验依赖程度高的情况，提出了人工神经网络应用于磁场匀场的手段。通过基于Matlab的神经网络设计，建立了一个匀场的BP网络模型。由于在磁体的不同的均匀度条件下匀场的方法不同，所以该网络由两个子网络组成----粗调网络和精调网络，并分别对其进行训练。该模型建立之后，可以快速、准确地匀场，使无源匀场不再盲目得进行多次尝试，一次就能使判断准确。通过人工神经网络方法匀场的磁体，均匀度高，满足了核磁共振岩心分析仪的匀场环境。     

人造碱基与人工合成生命

文章就人造碱基及相关人工合成生命领域内的代表性研究成果进行了全景式综述,着重介绍了氢键互补及疏水性两种最为主要的人造碱基对,探讨了具有里程碑意义的六核酸分子人工合成生命及其对合成生物学未来发展的影响,简单综述了人造碱基在DNA检测诊断试剂盒等方面的成功应用实例。最后对这一研究领域的前景和面临的挑战进行了展望。     

基于贝叶斯正则化BP神经网络的上市公司信用评价研究

在神经网络训练中引入贝叶斯正则化以提高模型的推广能力。通过与未使用提高推广能力方法和使用提前停止法两种情况进行比较,表明基于贝叶斯正则化的BP神经网络模型在上市公司信用评价中的推广能力有显著的提高。