基于快速神经网络架构搜索的鲁棒图像水印网络算法

为解决深度学习在图像水印算法中计算量大且模型冗余的问题,提高图像水印算法在抵抗噪声、旋转和剪裁等攻击时的鲁棒性,提出基于快速神经网络架构搜索（neural architecture search,NAS）的鲁棒图像水印网络算法。通过多项式分布学习快速神经网络架构搜索算法,在预设的搜索空间中搜索最优网络结构,进行图像水印的高效嵌入与鲁棒提取。首先,将子网络中线性连接的全卷积层设置为独立的神经单元结构,并参数化表示结构单元内节点的连接,预先设定结构单元内每个神经元操作的搜索空间;其次,在完成一个批次的数据集训练后,依据神经元操作中的被采样次数和平均损失函数值动态更新概率;最后,重新训练搜索完成的网络。水印网络模型的参数量较原始网络模型缩减了92%以上,大大缩短了模型训练时间。由于搜索得到的网络结构更为紧凑,本文算法具有较高的时间性能和较好的实验效果,在隐藏图像时,对空域信息的依赖比原始网络更少。对改进前后的2个网络进行了大量鲁棒性实验,对比发现,本文算法在CIFAR-10数据集上对抵抗椒盐噪声和旋转、移除像素行（列）等攻击优势显著;在ImageNet数据集上对抵抗椒盐高斯噪声、旋转、中值滤波、高斯滤波、JPEG压缩、裁剪等攻击优势显著,特别是对随机移除行（列）和椒盐噪声有较强的鲁棒性。     

面向离散文本舆情分析的分聚类方案

离散文本已经成为一种占据重要地位的舆情信息表现形式,根据离散文本的特点,提出基于特征概念网的离散文本舆情信息的分聚类框架,在此基础上给出分聚类方案。在聚类算法中,运用了遗传算法的全局并行搜索能力、k—means的高效局部聚类能力和小生境的保持种群多样性抑制漂移能力;在分类算法中,先将训练文本库进行类内聚类成子类,对子类构建特征概念网以生成替代该子类的文本,再用KNN算法进行分类。最后结合舆情分析进一步提出了可用的改进方案。     

基于影响力形成的论坛意见领袖识别

为了在网络论坛中提取意见领袖,文中通过抽象模拟论坛舆论意见领袖影响力的形成过程,建立了影响力向量模型,进而提出了可以利用影响力向量的模和方向这对参数来筛选意见领袖,并以天涯论坛某日的一个话题为研究个案,阐述了该参数的选择依据。再通过聚类分析比较不同类型的论坛舆论参与者的特点,将分析结果与提出模型筛选的结果相比对,论证文中方法结果的有效性和准确性,反应出文中提出参数的优越性。     

非常规突发事件网络舆情热度评价体系构建——基于BP神经网络算法

Because unexpected emergency owns the characteristics of explosive, uncertain evolution direction and group diffusion, more and more researchers concentrate on and try to control it. In addition, considering the force of network, the information of the unexpected emergency will be spread and enlarged rapidly on internet. It is a new viewpoint using the indicator system to estimate the heat degree of net-mediated public opinion on unexpected emergency, which can reveal the underlying reasons about the formation of the heat degree. Moreover, we use BP (Back Propagation) neural network method instead of traditional subjective weight assignment to calculate the weights of the indicators which can make evaluation results more accurate and objective.     

Epistemic Communities under Active Inference

The spread of ideas is a fundamental concern of today’s news ecology. Understanding the dynamics of the spread of information and its co-option by interested parties is of critical importance. Research on this topic has shown that individuals tend to cluster in echo-chambers and are driven by confirmation bias. In this paper, we leverage the active inference framework to provide an in silico model of confirmation bias and its effect on echo-chamber formation. We build a model based on active inference, where agents tend to sample information in order to justify their own view of reality, which eventually leads to them to have a high degree of certainty about their own beliefs. We show that, once agents have reached a certain level of certainty about their beliefs, it becomes very difficult to get them to change their views. This system of self-confirming beliefs is upheld and reinforced by the evolving relationship between an agent’s beliefs and observations, which over time will continue to provide evidence for their ingrained ideas about the world. The epistemic communities that are consolidated by these shared beliefs, in turn, tend to produce perceptions of reality that reinforce those shared beliefs. We provide an active inference account of this community formation mechanism. We postulate that agents are driven by the epistemic value that they obtain from sampling or observing the behaviours of other agents. Inspired by digital social networks like Twitter, we build a generative model in which agents generate observable social claims or posts (e.g., ‘tweets’) while reading the socially observable claims of other agents that lend support to one of two mutually exclusive abstract topics. Agents can choose which other agent they pay attention to at each timestep, and crucially who they attend to and what they choose to read influences their beliefs about the world. Agents also assess their local network’s perspective, influencing which kinds of posts they expect to see other agents making. The model was built and simulated using the freely available Python package pymdp. The proposed active inference model can reproduce the formation of echo-chambers over social networks, and gives us insight into the cognitive processes that lead to this phenomenon.     

A stochastic model for the HIV epidemic in homosexual populations involving age and race

In this paper, we derive a stochastic model for the HIV epidemic in homosexual populations involving age and race. To account for effects of different mixing patterns, a low risk selection rule is introduced. The model is then formulated in terms of chain multinomial distributions by means of which the means are derived. Some simulation studies by computer indicate clearly that age and race have a significant impact on the HIV epidemic.     

复杂网络上的意见动力学对谣言传播的影响

通过在SIR(susceptible-infected-recovered)模型中引入抑制者对谣言的辟谣机制研究了在线社交网络上的意见动力学对谣言传播的影响.在这一模型中,节点可以与自身的邻居组成1个群,传播者可以通过该群传播信息,抑制者也可以在此群中对信息发表意见进行辟谣.辟谣机制在降低未知者对于谣言的接受概率的同时也可以促使传播者向抑制者转变.本文采用ER(Erd?s-Rényi)随机网络、无标度网络以及真实的社交网络研究了抑制者的沉默概率对于谣言传播范围的影响.首先发现,谣言传播的过程以传播者的峰值为界可以分为两个阶段,即谣言自由传播的前期以及抑制者和传播者互相制衡的后期;其次,谣言的传播会随着抑制者的沉默概率的增大而突然暴发.在谣言暴发阈值之下,沉默概率的增大不会导致谣言传播范围显著增大,这是由于未知者在感知到谣言并转变为传播者后又迅速转变为抑制者;而当沉默概率达到谣言暴发阈值时,抑制者将不能控制传播者对谣言的传播从而导致抑制者的降低和谣言的暴发;最后,无标度上的谣言自由传播的前期阶段比随机网络持续的时间更短,从而使无标度上的谣言更难以暴发.本文的模型综合考虑了意见动力学和谣言传播的相互作用,更加真实地模拟了真实世界社交网络中的谣言传播过程.为谣言传播的控制和干预提供了一些有用的思路和见解.     

Compact Matrix Expressions for Generalized Wald Tests of Equality of Moment Vectors

Asymptotic chi-squared test statistics for testing the equality of moment vectors are developed. The test statistics proposed are generalized Wald test statistics that specialize for different settings by inserting an appropriate asymptotic variance matrix of sample moments. Scaled test statistics are also considered for dealing with nonstandard conditions. The specialization will be carried out for testing the equality of multinomial populations, and the equality of variance and correlation matrices for both normal and nonnormal data. When testing the equality of correlation matrices, a scaled version of the normal theory chi-squared statistic is proven to be an asymptotically exact chi-squared statistic in the case of elliptical data.     

基于微博的网络舆情管控技术研究

微博作为社交网络发展的最新产物,已逐渐成为网络舆情爆发的始发地。微博舆情管控有效的措施是发布舆论疏导信息进行正确导向。研究基于微博网站的开放平台,通过注册合法用户,采取群发消息、群发评论、群发@信息等技术途径,增强舆论导向信息传播的速度和密度、信息投送的精度和力度。最后通过国内主流的微博网站开放平台搭建实验环境,对微博舆情管控技术进行了验证,能够较好地实现微博舆情疏导信息形成几何数的传播速度。     

基于社交网络的舆情安全形势分析

社交网络在为广大网民提供自由交流的同时,也成为网络舆情的主要发源地。文中从网络舆情的产生和传播特点出发,对目前中国网络信息安全形式进行了深入分析,介绍了网络突发事件、热点事件在舆情产生中的作用,指出了正确引导舆情的策略方法。分析认为,网络舆情关系到国家安全和社会稳定,提高网络国防意识,保卫网络边疆,对网络舆情进行有效监控和正确引导刻不容缓。