针对堆石坝填筑进度控制以及土石方动态调运问题,受AlphaGo-Zero的启发,本文提出了一个基于蒙特卡洛树搜索(Monte Carlo tree search,MCTS)的土石方智能动态调配模型。该模型以当前累计填筑工程量、紧邻前一月份完成工程量以及当前月份...针对堆石坝填筑进度控制以及土石方动态调运问题,受AlphaGo-Zero的启发,本文提出了一个基于蒙特卡洛树搜索(Monte Carlo tree search,MCTS)的土石方智能动态调配模型。该模型以当前累计填筑工程量、紧邻前一月份完成工程量以及当前月份为状态,用各月填筑工作面对应的填筑可达强度约束动作空间,综合考虑节点工期、总工期、坝面施工机械费用和土石方调运费用等因素构造奖励函数。此外,结合本文研究问题的特点,对MCTS迭代中的上限置信区间算法(upper confidence bound apply to tree,UCT)进行了改进和比较分析,最后以一个工程实例对本文提出模型的有效性进行了验证分析。结果表明,与施工仿真相比,以MCTS为框架的土石方动态调配模型的计算分析时间大大减少,为土石方动态调配问题提供了新的模型与手段。展开更多
作为一种典型的欺骗防御手段,蜜罐技术在主动诱捕攻击者方面具有重要意义。然而现有设计方法主要通过博弈模型来优化蜜罐的诱捕决策,忽略了攻击者的信念对双方博弈决策的影响,存在自适应优化决策能力弱、易被攻击者识破并利用等不足。为...作为一种典型的欺骗防御手段,蜜罐技术在主动诱捕攻击者方面具有重要意义。然而现有设计方法主要通过博弈模型来优化蜜罐的诱捕决策,忽略了攻击者的信念对双方博弈决策的影响,存在自适应优化决策能力弱、易被攻击者识破并利用等不足。为此,提出了基于信念的蜜罐博弈机制(Belief Based Honeypot Game Mechanism, BHGM)。BHGM基于攻击者完成任务的多轮博弈过程,重点关注蜜罐采取动作对攻击者信念的影响以及信念对攻击者是否继续攻击的影响。同时,基于树上限置信区间(Upper Confidence Bound Apply to Tree, UCT)设计了信念驱动的攻防最优策略求解算法。仿真实验结果表明,信念驱动的攻击方策略能基于当前信念选择继续攻击或及时止损以获得最大收益,而信念驱动的蜜罐策略在考虑风险的情况下能尽量降低攻击方怀疑,以诱骗其继续攻击,从而获得更大收益。展开更多
Introduction:School-aged children are primary vectors for influenza transmission through their frequent close contact in educational settings and developing immune awareness.Since 2019,the Shenzhen municipal governmen...Introduction:School-aged children are primary vectors for influenza transmission through their frequent close contact in educational settings and developing immune awareness.Since 2019,the Shenzhen municipal government has implemented annual,free,influenza vaccination programs targeting eligible primary and secondary school students.However,evidence-based strategies specifically tailored to this demographic remain insufficient.Methods:This study analyzed weekly influenza-like illness(ILI)surveillance data and laboratory-confirmed positivity rates from Shenzhen during the 2023–2024 season.It developed an age-stratified Susceptible–Exposed–Symptomatic–Asymptomatic–Recovered–Hospitalized–Deceased–Vaccinated compartmental model integrated with the Ensemble Adjustment Kalman Filter(EAKF)algorithm to estimate historical transmission parameters and quantify vaccination impact.The Upper Confidence Bound applied to Trees(UCT)algorithm was used to optimize the vaccination schedule and evaluate multiple strategic scenarios comparatively.Results:Compared to a no-vaccination scenario,the current government strategy prevented approximately 1,285,925[95%confidence interval(CI):1,240,671–1,331,180]symptomatic infections and 56,956(95%CI:55,118–58,793)hospitalizations.Under identical vaccine supply conditions,the optimized strategy recommends vaccinating 30%,25%,and 5%of school-aged children in November,December,and January,respectively.This optimized approach would avert approximately 1,469,368(95%CI:1,392,734–1,546,002)symptomatic infections and 64,442(95%CI:61,269–67,615)hospitalizations—representing 14.3%and 13.1%improvements over the government strategy,respectively.Additionally,a generic strategy developed using 2017–2019 data performed well during 2023–2024,demonstrating cross-seasonal adaptability.Conclusions:Concentrating influenza vaccination efforts among school-enrolled children during November and December significantly reduces disease burden and represents a critical strategy for controlling influenza transmission.展开更多
文摘针对堆石坝填筑进度控制以及土石方动态调运问题,受AlphaGo-Zero的启发,本文提出了一个基于蒙特卡洛树搜索(Monte Carlo tree search,MCTS)的土石方智能动态调配模型。该模型以当前累计填筑工程量、紧邻前一月份完成工程量以及当前月份为状态,用各月填筑工作面对应的填筑可达强度约束动作空间,综合考虑节点工期、总工期、坝面施工机械费用和土石方调运费用等因素构造奖励函数。此外,结合本文研究问题的特点,对MCTS迭代中的上限置信区间算法(upper confidence bound apply to tree,UCT)进行了改进和比较分析,最后以一个工程实例对本文提出模型的有效性进行了验证分析。结果表明,与施工仿真相比,以MCTS为框架的土石方动态调配模型的计算分析时间大大减少,为土石方动态调配问题提供了新的模型与手段。
文摘作为一种典型的欺骗防御手段,蜜罐技术在主动诱捕攻击者方面具有重要意义。然而现有设计方法主要通过博弈模型来优化蜜罐的诱捕决策,忽略了攻击者的信念对双方博弈决策的影响,存在自适应优化决策能力弱、易被攻击者识破并利用等不足。为此,提出了基于信念的蜜罐博弈机制(Belief Based Honeypot Game Mechanism, BHGM)。BHGM基于攻击者完成任务的多轮博弈过程,重点关注蜜罐采取动作对攻击者信念的影响以及信念对攻击者是否继续攻击的影响。同时,基于树上限置信区间(Upper Confidence Bound Apply to Tree, UCT)设计了信念驱动的攻防最优策略求解算法。仿真实验结果表明,信念驱动的攻击方策略能基于当前信念选择继续攻击或及时止损以获得最大收益,而信念驱动的蜜罐策略在考虑风险的情况下能尽量降低攻击方怀疑,以诱骗其继续攻击,从而获得更大收益。
基金Supported by the National Key R&D Program of China(2022YFE0112300,2023YFC2308701)the National Natural Science Foundation of China(82304204)+5 种基金the Natural Science Foundation of Guangdong Province(2025A1515011908)the Technological Innovation Team of Shaanxi Province(2025RS-CXTD-009)the International Cooperation Project of Shaanxi Province(2025GH-YBXM-017)the Shenzhen-Hong Kong-Macao Science and Technology Project(Category C)(SGDX20230821091559022)the Fundamental Research Funds for the Central Universities(G2024WD0151,D5000240309)the CAMS Innovation Fund for Medical Sciences(2021-I2M-1-044).
文摘Introduction:School-aged children are primary vectors for influenza transmission through their frequent close contact in educational settings and developing immune awareness.Since 2019,the Shenzhen municipal government has implemented annual,free,influenza vaccination programs targeting eligible primary and secondary school students.However,evidence-based strategies specifically tailored to this demographic remain insufficient.Methods:This study analyzed weekly influenza-like illness(ILI)surveillance data and laboratory-confirmed positivity rates from Shenzhen during the 2023–2024 season.It developed an age-stratified Susceptible–Exposed–Symptomatic–Asymptomatic–Recovered–Hospitalized–Deceased–Vaccinated compartmental model integrated with the Ensemble Adjustment Kalman Filter(EAKF)algorithm to estimate historical transmission parameters and quantify vaccination impact.The Upper Confidence Bound applied to Trees(UCT)algorithm was used to optimize the vaccination schedule and evaluate multiple strategic scenarios comparatively.Results:Compared to a no-vaccination scenario,the current government strategy prevented approximately 1,285,925[95%confidence interval(CI):1,240,671–1,331,180]symptomatic infections and 56,956(95%CI:55,118–58,793)hospitalizations.Under identical vaccine supply conditions,the optimized strategy recommends vaccinating 30%,25%,and 5%of school-aged children in November,December,and January,respectively.This optimized approach would avert approximately 1,469,368(95%CI:1,392,734–1,546,002)symptomatic infections and 64,442(95%CI:61,269–67,615)hospitalizations—representing 14.3%and 13.1%improvements over the government strategy,respectively.Additionally,a generic strategy developed using 2017–2019 data performed well during 2023–2024,demonstrating cross-seasonal adaptability.Conclusions:Concentrating influenza vaccination efforts among school-enrolled children during November and December significantly reduces disease burden and represents a critical strategy for controlling influenza transmission.
