一种改进的自组织映射算法求解旅行商问题

目前,没有求解旅行商问题的非常有效的方法。提出了一种求解该问题的LNSOM算法,在自组织映射算法的基础上,改进了学习率和邻域函数变量。利用matlab 2011软件进行求解,其中5个旅行商问题实例的结果优于MSTSP和SETSP算法,另外,10个实例的平均误差为1.445 6%。实验结果表明,新算法的误差更小,并保持了SOM算法较低的计算复杂度。     

自适应交配限制概率的自组织多目标演化算法

为平衡多目标演化算法求解不同优化问题以及求解同一优化问题时不同搜索阶段的勘探与开采能力,并考虑到减小聚类算法辅助演化算法时产生的计算开销,提出了一种基于自适应交配限制概率的自组织多目标演化算法（adaptive mating restriction probability based self-organizing multiobjective evolutionary algorithm, ASMEA）.首先,ASMEA在每一代利用自组织映射（self-organizing map, SOM）算法建立了演化种群个体间的邻居关系,基于此关系有利于算子实施恰当的重组操作,并在演化算法后期产生优质解,与此同时,为了节省利用SOM建立当前种群个体之间的邻居关系时引起的计算开销,将SOM与演化算法相融合,交替地进行SOM训练与种群演化.然后,运用交配限制概率控制交配父代来源于SOM发现的邻居种群或者是整个种群,以分别加强开采和勘探. 最后,根据采用不同父代来源的重组在过去一定代数产生后代个体的效用,自适应地调整算法的交配限制概率. 利用ASMEA和5种具有代表性的多目标演化算法对标准测试题进行求解,求解结果表明:ASMEA在搜索质量、搜索效率以及可视化方面优于其他5种算法,从而验证了ASMEA算法对多目标优化问题具有良好的求解性能.     

图像数据库分类的自组织技术

本文采用 Kohoenmap自组织神经网络技术 ,提出一种依据检索特征对图像数据库进行预分类的方法 .在文中利用这种预分类实现按类别浏览图像数据库 (image browsing by category)的功能 ,提出一种层次化的类 SS- tree的二分索引结构 .实现建立在预分类基础上的示例查询 .这种查询方法与穷搜索方法相比 ,具有更高的检索效率 ,检索次数随着库规模的增加不会显著增多 .基于图像库的实验验证了分类算法和检索算法的有效性     

基于自组织神经网络和模糊聚类的校园无线网用户日常行为分析

针对如何分析校园无线网络数据,挖掘数据中蕴藏的学生行为,更好地辅助教学管理,本文提出了在Hadoop平台构建基于自组织神经网络(SOFM)的模糊C-均值(FCM)聚类算法。该算法采用自组织神经网络与模糊C-均值聚类算法相结合,避免了模糊C-均值聚类算法初始化不当带来的误差,目标函数中采用马氏距离,自适应的调整了数据的几何分布。考虑到无线用户数据规模庞大,采用了Hadoop平台并行运行聚类算法。实验结果表明,本文提出的算法提高了聚类结果的准确性,有效地降低了时间复杂度,分析平台为学校管理层快速有效的做出决策提供了依据,研究分析方法对其它高校有较大地参考价值。     

一种协同半监督分类算法Co-S3OM

为了提高半监督分类的有效性, 提出了一种基于SOM神经网络和协同训练的半监督分类算法Co-S3OM (coordination semi-supervised SOM)。将有限的有标记样本分为无重复的三个均等的训练集, 分别使用改进的监督SSOM算法(supervised SOM)训练三个单分类器, 通过三个单分类器共同投票的方法挖掘未标记样本中的隐含信息, 扩大有标记样本的数量, 依次扩充单分类器训练集, 生成最终的分类器。最后选取UCI数据集进行实验, 结果表明Co-S3OM具有较高的标记率和分类率。     

基于K-means算法改进的SOM神经网络调制识别分类器

通信过程中,获得情报信息的关键步骤是清楚接收到的调制信号的调制方式.随着现代通信技术的高速发展,人工智能广泛应用于调制方式识别领域.提出将自组织特征映射(Self-Organizing feature Map,简称SOM网络)神经网络用于调制制式的识别.用K均值(K-means)聚类算法来寻找每类特征参数的两个聚类中心...     

基于松散偏好规则的群体机器人系统自组织协作围捕

针对群体机器人协作围捕,提出了一种基于松散偏好规则的自组织方法.首 先给出了个体机器人的自由运动模型和围捕行为的数学描述.通过对围捕行为的分解,构造松散偏好 规则来使个体机器人在自组织运动过程中相互协调最终形成理想的围捕队形.在此基础上,设计了个体自组织运动控制器.最后运用Lyapunov稳定性定理证明系统的稳定性.仿真和实验结果表 明,本文给出的自组织方法对于群体机器人协作围捕是行之有效的.     

Comparison between self-organizing map and principal component analysis for water quality assessment and hydro-geochemical characterization in dyke intruded complex geological settings

Hydro-geochemical characterization is challenging in dyke intruded complex geological setting. The comparison between self-organizing map (SOM) classification and principal component analysis (PCA) is used for better understanding of hydrogeological process surrounding Amarpur dyke in Dhanbad district, Jharkhand. Total 30 water samples were collected and tested for 12 physicochemical parameters. The K-means clustering with SOM grouped the water quality data into cluster 1 (46.67%, low mineralization), cluster 2 (36.67%, moderate mineralization) and cluster 3 (16.67%, high mineralization). The clusters of the majority of samples identified by PCA analysis is almost same as identified by SOM with little difficulty in discriminating between cluster 2 and cluster 3. The transformation of Ca-HCO₃ to Ca-Cl-SO₄ occurred because of exchange of Ca²⁺ with Na⁺ adsorbed in the aquifer leading excess of sulphate ions. The results of this study suggest that SOM is an effective tool for a better understanding of patterns and processes driving water quality.     

An Enhanced Group Key-Based Security Protocol to Protect 5G SON Against FBS

Network operators are attempting many innovations and changes in 5G using self-organizing networks (SON). The SON operates on the measurement reports (MR), which are obtained from user equipment (UE) and secured against malware and userspace programs. However, the synchronization signal block that the UE relies on to measure the wireless environment configured by a base station is not authenticated. As a result, the UE will likely gauge the wrong wireless environment configured by a false base station (FBS) and transmit the corresponding MR to the serving base station, which poisons the data used for 5G SONs. Therefore, the serving base stations must verify the authenticity of the MR. The 3GPP has advocated numerous solutions for this issue, including the use of public key certificates, identity-based keys, and group keys. Although the solution leveraging group keys have better efficiency and practicality than the other two, they are vulnerable to security threats caused by key leaks via insiders or malicious UE. In this paper, we analyze these security issues and propose an improved group key protocol that uses a new network function, called a broadcast message authentication network function (BMANF), which validates broadcasted messages on behalf of the UE. The protocol operates in two phases: initial and verification. During the initial phase, the 5G core network distributes a shared secret key to the BMANF and UE, allowing the latter to request an authentication ticket from the former. During the verification phase, the UE requests the BMANF to validate the broadcasted messages received from base stations using the ticket and its corresponding shared key. For evaluation, we formally verified the proposed protocol, which was then compared with alternative methods in terms of computing cost. As a result, the proposed protocol fulfills the security requirements and shows a lower overhead than the alternatives.     

Multiple input self-organizing-map ResNet model for optimization of petroleum refinery conversion units

This work introduces a deep-learning network, i.e., multi-input self-organizing-map ResNet (MISR), for modeling refining units comprised of two reactors and a separation train. The model is comprised of self-organizing-map and the neural network parts. The self-organizing-map part maps the input data into multiple two-dimensional planes and sends them to the neural network part. In the neural network part, residual blocks enhance the convergence and accuracy, ensuring that the structure will not be overfitted easily. Development of the MISR model of hydrocracking unit also benefits from the utilization of prior knowledge of the importance of the input variables for predicting properties of the products. The results show that the proposed MISR structure predicts more accurately the product yields and properties than the previously introduced self-organizing-map convolutional neural network model, thus leading to more accurate optimization of the hydrocracker operation. Moreover, the MISR model has smoother error convergence than the previous model. Optimal operating conditions have been determined via multi-round-particle-swarm and differential evolution algorithms. Numerical experiments show that the MISR model is suitable for modeling nonlinear conversion units which are often encountered in refining and petrochemical plants.