Self-Organising Maps as a Relevance Feedback Technique in Content-Based Image Retrieval

Self-Organising Maps (SOMs) can be used in implementing a powerful relevance feedback mechanism for Content-Based Image Retrieval (CBIR). This paper introduces the PicSOM CBIR system, and describes the use of SOMs as a relevance feedback technique in it. The technique is based on the SOM’s inherent property of topology-preserving mapping from a high-dimensional feature space to a two-dimensional grid of artificial neurons. On this grid similar images are mapped in nearby locations. As image similarity must, in unannotated databases, be based on low-level visual features, the similarity of images is dependent on the feature extraction scheme used. Therefore, in PicSOM there exists a separate tree-structured SOM for each different feature type. The incorporation of the relevance feedback and the combination of the outputs from the SOMs are performed as two successive processing steps. The proposed relevance feedback technique is described, analysed qualitatively, and visualised in the paper. Also, its performance is compared with a reference method.     

自适应类增量学习的物联网入侵检测系统

传统物联网入侵检测系统难以实时检测新类别攻击,为此,提出一种基于堆叠稀疏自编码器（SSAE）和自组织增量神经网络（SOINN）的物联网入侵检测方法。SSAE对已知类别的攻击样本进行稀疏编码和特征提取,所提取的特征输入SOINN,SOINN形成符合流量特征空间分布的拓扑结构。当出现新类别训练样本的特征时,SOINN自适应地生成新节点以建立新的局部拓扑结构。为了保留SSAE在旧类别样本上的知识,对SOINN已有的拓扑结构施加约束,通过误差反向传递间接约束SSAE权重的变化。针对SOINN在新类别上产生的新局部拓扑结构进行自适应聚合和优化,从而实现新类别样本学习。实验结果表明,该方法能基于新类别数据对模型进行增量训练而无需历史类别数据,在CIC-IDS2017数据集的动态数据流中能有效检测新类别攻击同时缓解旧类别数据中存在的灾难性遗忘问题,在初始已知数据集上的准确率达到98.15%,完成3个阶段的类别增量学习后整体准确率仍能达到57.34%,优于KNN-SVM、mCNN等增量学习方法。     

基于Google Maps API的车辆监控管理系统设计与实现

车辆实时监控是车队现代化管理的重要手段,能有效提高车队的生产力。本文基于CoogleMapsAPI和Ajax技术设计车辆监控平台和数据通信服务,涉及GPS数据采集服务、协议解码、实时车辆定位、轨迹回放、围栏设置等功能的设计和实现方法,为车辆监控平台的设计提供参考。     

基于小波对比度和神经网络的图像隐写方法

为使通信安全在传输过程中提供较大的秘密信息嵌入量,并保持较好的载密图像质量,提出一种基于自组织特征映射神经网络和小波对比度的图像隐写方法。将载体图像分成固定大小的小块,采用小波一级分解并计算其小波对比度,利用自组织特征映射神经网络将小块分为3类,采用模算子技术嵌入秘密信息。实验结果表明,该方法有较大的嵌入量并保持良好的载密图像质量。     

In search of better route guidance instructions

Safety concerns have prompted designers of in-vehicle route guidance and information systems (IVRGIS) to make more use of an ‘audible interface’ to convey guidance instructions. Previous research has shown that, contrary to expectations, detailed guidance instructions can have a detrimental effect upon wayfinding performance, particularly for elderly individuals. In response to these findings a second series of experiments was carried out to try to improve the effectiveness of route guidance. Using the same procedure, 40 male and 40 female drivers aged 18–35 years watched video footage of journeys through an unfamiliar area, while hearing guidance that linked direction instructions to landmarks visible at the decision point. Results showed that those who heard these amended instructions performed significantly better than other groups at a wide range of tasks designed to measure the spatial knowledge they had acquired. This group performed better than groups who heard nothing, and groups who heard full guidance. The results support the view that, whereas full guidance instructions can have a negative impact upon wayfinding performance, less complex instructions that link landmarks to directions have the capacity to enhance wayfinding performance. This is because instructions of this form strengthen the associations made between directions to be taken and the spatial location of these turns, resulting in the formation of a strong representation of the route. The results also showed that those who had been driving for 1 year or less were significantly less accurate at these tasks than those with more driving experience. This result has important implications for the widespread implementation and use of IVRGIS: suggesting that, for newly qualified drivers, who have not yet developed the ability to attend to and process information while controlling the vehicle, attending to route guidance instructions might have a detrimental effect upon driver safety.     

基于四元数和SOM神经网络的彩色图像边缘检测

针对传统边缘检测方法无法考虑到彩色图像各颜色分量的相关性,以及边缘提取效果受阈值影响的不足,提出一种四元数和自组织神经网络(SOM)相结合的彩色图像边缘检测算法。根据四元数柯西积分公式和四元数矢量积性质,构造图像的边缘特征向量对SOM神经网络进行训练,然后用训练好的SOM网络提取边缘。实验表明,该方法具有良好的边缘检测效果,并有较强的细节保持能力。     

农产品价格信息采集与预警系统设计与实现

为了实现农产品价格信息采集发布的及时性和有效性,提高农产品数据质量,提出一种农产品信息采集与预警系统的设计方法.通过运用Google Maps提供的应用程序开发接口和ASP.NET技术,并集成移动通讯技术和智能客户端技术,设计了农产品信息的采集、处理、发布,系统数据的预警以及基础数据维护等功能模块,保证了系统的实用性、兼容性和可扩展性.实验结果表明,在农产品价格信息采集方面,该系统实现了农产品价格信息的实时采集、处理、发布及预警功能,保证了数据的有效性和准确性,解决了农产品价格信息采集的滞后性等问题,有利于农产品信息的快速交流.     

SOM网络在透平机械状态监测中的应用

讨论了SOM网络的可视化技巧,结合该技巧研究了SOM网络在透平机械状态监测中的应用。仿真分析和实例研究表明,应用SOM网络能较好地识别出透平机械的不同状态,并及时监测到透平机械状态发生的变化。     

基于Google Maps API的高速公路WebGIS设计与应用

设计了基于Google Maps API技术的WebGIS系统,同时讨论了Google Maps的定制方法和功能设计,给出了定制设计的部分代码。文章结合高速公路WebGIS的设计实例,给出了整体系统示意图并对系统的各项功能做了详细介绍。     

The Role of Cadastral Surveys and Maps in Land Settlement from England

The role of surveying and mapping the landscapes of colonial settlement is discussed. Cadastral maps (maps of landed property) have been used since the end of the Middle Ages by individuals and institutions to establish title to land. From the 17th century, cadastral maps were used by state governments in both the Old and New Worlds to organize, control and record the settlement of 'empty' landscapes. Maps were the instruments that enabled the settlement ideals of colonial governments to be realized. These ranged from the encouragement of large plantations, as on the southern seaboard of North America, to individual proprietorship of holdings disposed with the regularity of the grid, as with the federal land disposals of the USA. They also encompassed the strict limitation of land availability in order to establish a capitalist society with farmers and wage labourers, as in the Wakefield settlements of Australia and New Zealand.