概率级联布尔网络的集镇定及其应用

随着系统生物学和医学的迅速发展,基因调控网络已经成为一个热点研究领域.布尔网络作为研究生物系统和基因调控网络的一种重要模型,近年来引起了包括生物学家和系统科学家在内的很多学者的广泛关注.本文利用代数状态空间方法,研究了概率级联布尔网络的集镇定问题.首先给出概率级联布尔网络集镇定的定义,并利用矩阵的半张量积给出了概率级联布尔网络的代数表示.其次基于该代数表示,定义了一组合适的概率能达集,并给出了概率级联布尔网络集镇定问题可解的充要条件.最后将所得的理论结果应用于概率级联布尔网络的同步分析及n人随机级联演化布尔博弈的策略一致演化行为分析.     

基于多策略融合粒子群的无人机对地攻击模糊博弈决策

针对无人机协同对地攻击的复杂性和不确定性,联合防空火力压制与对地目标打击任务,引入存活因子、摩擦因子和状态因子等概念,考虑目标威胁度的模糊性,结合生存概率和武器消耗等因素,建立一种多阶段的模糊多目标任务分配规划模型.为更好地描述攻击任务的对抗性和多策略性,以博弈论为框架,将规划模型转化为模糊多目标双矩阵博弈综合集结模型.利用必要性理论将集结模型中的不确定性目标清晰化处理,进而运用熵权法对多个目标进行加权求和,将其转化为单目标双矩阵博弈模型.提出基于多策略融合粒子群算法的纳什均衡求解方法,通过引入自适应惯性权重、动态反向学习与局部变异策略,在增强种群多样性的同时,保证粒子群局部精确搜索能力.算例仿真结果验证了所提模型和方法的有效性.     

基于机器视觉的中国象棋人机博弈系统设计

通过对图像处理及人机博弈相关理论的研究,实现了一种中国象棋人机博弈机器人系统的设计。该系统通过摄像头采集棋盘图像,之后将图像传入计算机,同时计算机通过图像剪切方法和基于对数极坐标变换&傅里叶变换的模板匹配方法实现棋盘中棋子的定位与识别,之后调用中国象棋人机博弈算法得出下一步应进行的走法,计算机通过串口将走法信息传入下位机单片机中,单片机根据走法信息控制机械臂实现棋子的移动,完成与人的对弈过程。经测试,该系统能够完成基本的功能并且具有较好的稳定性。     

Improving Data Utility Through Game Theory in Personalized Differential Privacy

Journal of Computer Science and Technology - Due to dramatically increasing information published in social networks, privacy issues have given rise to public concerns. Although the presence of...     

ICOMDT:一个面向动态任务的交互计算模型

近年来,包含动态任务的交互式系统得到了广泛的应用.基于现有对用户与动态任务交互的研究,提出一个面向动态任务的定量化可计算的交互模型ICOM^DT,用于解释用户与动态任务的交互行为,并实现用户意图预测.更具体地,将ICOM^DT应用于运动目标选择任务,设计了两个实验以验证模型的有效性.实验1收集用户数据对模型进行拟合并预测用户选择的错误率,实验结果表明,能够很好地拟合且预测值也与真实值接近;实验2将模型对用户意图的理解拓展为一种辅助运动目标选择技术ICOMPointer,通过在具体的游戏场景中与Windows原本的选择技术及两种现有的增强目标选择技术对比,发现使用ICOMPointer选中目标的平均完成时间更短,完成速度更快.ICOM^DT模型的提出对计算机理解用户意图,提高用户与动态任务的交互效率有着重要意义.     

Effects of control-display gain and postural control method on distal pointing performance

The present study investigated the effects of different joint-based interactive methods and control-display gains on performance and usability for distant pointing by using ISO 9241–9 test procedures. Ten healthy participants performed multi-directional tapping tests with three joint-based interactive methods (wrist-, elbow-, and shoulder-based interactive methods) and three CD gains (0.6, 1.0, and 1.7) by using an air mouse as the input device. The experimental results indicated that the wrist-based interactive method is a better interactive method for tasks where pointing speed is the highest priority and the 1.7 CD gain, which was found to cause significantly higher MT and ER and lower TP and subjective usability rating scores than wrist- and elbow-based pointing, should be avoided in distal pointing tasks. Furthermore, the best suggested CD gains for the wrist-based interactive method are 1.0 and 1.7. The recommended CD gain setting of the elbow-based interactive method is 0.6, and 1.0 CD gain is feasible for the shoulder-based interactive method.     

A novel matrix game with payoffs of Maxitive Belief Structure

Maxitive Belief Structures (MBSs), emerge as a novel class of belief structure to represent imprecise and uncertain information with a specific structure allowing it to model both probability nonspecificity and imprecise. A novel matrix game with payoffs of MBS is presented. The proposed model can be used to describe interactions among players, and an effective method is developed to find the solution of a game through reaching an equilibrium point. A numerical example is given to verify the validness of the proposed model.     

Alone or together? Exploring the role of desire for online group gaming in players’ social game addiction

Drawing on dual-system theory, this study examines the role of desire for online group gaming in online social game addiction. By conducting a longitudinal online survey, we show that attitude, anticipated enjoyment, group norm, and social identity contribute to the development of desire for online group gaming. Such desire facilitates habit and self-regulation deficiency, which, in turn, fosters online social game addiction. This study contributes to the literature by theorizing a nomological network of desire for online group gaming on online social game addiction and by offering implications for managing addictive gaming.     

势博弈与时变Log-linear分布式拓扑控制算法

针对UWSNs由网络拓扑控制不稳定、频繁变化引起网络的能耗不均衡、网络生存周期短等问题,从分析传感器节点受水下不确定性因素对UWSNs的拓扑结构演化入手,构建分布式水下传感器节点状态变量描述模型,归纳出节点之间和节点与环境之间多目标交互、协同、决策的UWSNs拓扑控制优化问题,将多目标优化问题映射成博弈论优化问题,再利用势博弈与Log-linear分布式学习规则实现博弈中节点策略行为的更新过程,采用非齐次马尔可夫链理论证明网络拓扑控制目标函数的优化问题收敛到最大化势博弈函数的解,从而达到保持网络均衡、延长网络生存周期的目的.     

一种在矩阵空间中识别文本蕴涵的动态交互网络

针对文本蕴涵问题提出一种动态交互网络（dynamic interactive network,DIN）进行识别。不同于已有交互模型,DIN将两句词向量投射到二维矩阵空间中进行交互,然后利用输出矩阵为同时处理上下文信息和控制信息流动的GRU编码器生成动态权重。前者通过更高阶形式的信息交互挖掘深层逻辑片段,后者通过改变交互信息与上下文信息的结合模式帮助编码器有效区分两者的重要性差异。模型在SNLI测试集上获得了88.0%的识别准确度,超过已有的最佳模型,且使用的训练参数仅为它的一半。