该研究提出了一种基于特征提取的深度神经网络(ResNet),用于对局灶性肝脏病变的超声波图像进行分类。这些优势使其能够有效地从超声图像中提取肝脏损伤的迹象,并实现准确的分类。基于包含血管瘤、脂肪肝、肝转移瘤、肝囊肿和正常肝脏的...该研究提出了一种基于特征提取的深度神经网络(ResNet),用于对局灶性肝脏病变的超声波图像进行分类。这些优势使其能够有效地从超声图像中提取肝脏损伤的迹象,并实现准确的分类。基于包含血管瘤、脂肪肝、肝转移瘤、肝囊肿和正常肝脏的超声图像数据集,该模型在实验验证中提供了出色的分类效果。在测试集上,该模型的准确率达到了93.99%。本文将该模型与AlexNet和VGGNet模型进行了比较,结果表明作者获得的模型分类效果更好,在准确率、参数数量和学习效率方面都有显著优势,并且具有很强的泛化能力。这项研究在病灶性肝脏病变的超声波图像分类任务中具有潜在的应用价值,可为临床医生提供准确、快速的辅助诊断工具。This study proposes a feature extraction-based deep neural network (ResNet) for classifying ultrasound images of focal liver lesions. These advantages enable it to effectively extract signs of liver damage from ultrasound images and achieve accurate classification. Based on an ultrasound image dataset containing hemangioma, fatty liver, liver metastases, liver cysts, and normal liver, the model provided excellent classification results in experimental validation. On the test set, the model achieved an accuracy of 93.99%. This article compares this model with AlexNet and VGGNet models. The results show that the model obtained by the author has better classification results, has significant advantages in accuracy, number of parameters, and learning efficiency, and has strong generalization ability. This study has potential application value in the task of ultrasonic image classification of focal liver lesions and can provide clinicians with an accurate and rapid auxiliary diagnostic tool.展开更多
本文采用自适应控制和事件触发策略研究了矩阵加权网络下多智能体系统的一致性问题。不同于传统的纯量加权网络,智能体及其邻居之间的通信由正定或半正定的矩阵刻画。利用矩阵权值将更有利于反映智能体之间的逻辑依赖关系。结合动态事...本文采用自适应控制和事件触发策略研究了矩阵加权网络下多智能体系统的一致性问题。不同于传统的纯量加权网络,智能体及其邻居之间的通信由正定或半正定的矩阵刻画。利用矩阵权值将更有利于反映智能体之间的逻辑依赖关系。结合动态事件触发策略设计了两种不同的自适应控制协议,即基于边的自适应协议和基于节点的自适应协议,这些控制协议具有灵活调整控制输入的性能。给出了实现一致性的充分条件,并分析了芝诺行为的排除。最后,通过数值仿真验证了理论分析的有效性。This paper endeavors to investigate the consensus problem for multi-agent systems under matrix-weighted networks by employing adaptive control and event-triggered strategies. Different from conventional scalar-weighted networks, the interconnections among agents and their neighbors are characterized by positive definite or positive semi-definite matrices. The utilization of matrix weights is more beneficial to reflect the logical inter-dependency among agents. Two different adaptive control protocols, namely edge-based adaptive protocol and node-based adaptive protocol, are designed in combination with dynamic event-triggered strategies. These protocols can facilitate the flexible manipulation of control inputs. Sufficient conditions for achieving consensus are provided and the exclusion of Zeno behavior is analyzed. Finally, the validity of the theoretical analysis is verified through numerical simulation.展开更多
乳腺癌是女性常见恶性肿瘤之一,严重威胁女性健康。在乳腺癌手术中,准确测量乳房和切除组织的体积对于手术规划和乳腺重建至关重要。本研究构建了一个基于深度学习的分割框架,用于乳房和切除组织的MRI体积分割。该框架不仅展现了良好的...乳腺癌是女性常见恶性肿瘤之一,严重威胁女性健康。在乳腺癌手术中,准确测量乳房和切除组织的体积对于手术规划和乳腺重建至关重要。本研究构建了一个基于深度学习的分割框架,用于乳房和切除组织的MRI体积分割。该框架不仅展现了良好的分割精度,还涵盖了偏置场矫正这一关键预处理步骤。我们收集并手动标注了一个包含47例患者的MRI数据集,这些数据涵盖了不同的年龄、乳房大小和四类采集参数。在交叉验证中,U-Net网络在全乳分割和切除组织分割任务中表现最佳,平均Dice系数分别为96.54和92.37。在测试集上,U-Net网络同样展现了优异的分割效果,平均Dice系数分别为94.23和84.53。实验结果表明,所提出的框架能够精准且高效地量化全乳体积和切除组织体积,为临床乳腺手术提供数据支持。Breast cancer is one of the most common malignant tumors in women, which seriously threatens women’s health. In breast cancer surgery, accurate measurement of breast and excised tissue volume is critical for surgical planning and breast reconstruction. This study developed a deep learning based segmentation framework for MRI volume segmentation of breasts and excised tissues. This framework not only demonstrates good segmentation accuracy, but also covers the key preprocessing step of bias field correction. We collected and manually annotated an MRI dataset containing 47 patients, covering different ages, breast sizes, and four types of acquisition parameters. In cross validation, the U-Net network performed the best in both whole milk segmentation and excised tissue segmentation tasks, with average Dice coefficients of 96.54 and 92.37, respectively. On the test set, the U-Net network also demonstrated excellent segmentation performance, with average Dice coefficients of 94.23 and 84.53, respectively. The experimental results indicate that the proposed framework can accurately and efficiently quantify total breast volume and excised tissue volume, providing data support for clinical breast surgery.展开更多
本文主要研究在月光型顶点算子代数中满足一定条件的2对Ising向量生成的顶点算子代数的结构,这2对Ising向量分别生成1个3A代数,并且生成的2个3A代数的交包含一个同构于L(4/5, 0)⊕L(4/5, 3)的子顶点算子代数,本文证明了其一共有3种可能...本文主要研究在月光型顶点算子代数中满足一定条件的2对Ising向量生成的顶点算子代数的结构,这2对Ising向量分别生成1个3A代数,并且生成的2个3A代数的交包含一个同构于L(4/5, 0)⊕L(4/5, 3)的子顶点算子代数,本文证明了其一共有3种可能的顶点算子代数结构。In this paper, we mainly study the vertex operator algebra generated by two pairs of Ising vectors in the moonshine type vertex operator algebra. These two pairs of Ising vectors each generate one 3A algebra, and the intersection of the two generated 3A algebras contains a subvertex operator subalgebra that is isomorphic to L(4/5, 0)⊕L(4/5, 3). We have shown that there are three possible structures of vertex operators algebraic.展开更多
本文基于国家统计局官网关于纺织行业2018年01月~2023年12月的工业生产者出厂价格指数(PPI)数据,运用时间序列分析方法构建了ARIMA(1,2,1)模型,并对该模型进行了深入的拟合分析。利用所建立的模型对2024年全年12个月的PPI月度数据进行预...本文基于国家统计局官网关于纺织行业2018年01月~2023年12月的工业生产者出厂价格指数(PPI)数据,运用时间序列分析方法构建了ARIMA(1,2,1)模型,并对该模型进行了深入的拟合分析。利用所建立的模型对2024年全年12个月的PPI月度数据进行预测,并将预测结果与实际值进行对比,结果显示预测值的相对误差远低于5%,充分验证了模型具有良好的拟合度和预测精度。此外,在进行未来预测时,需要充分考虑外部因素的影响,并对模型进行适时的调整和优化,以提高预测的准确性和可靠性。Based on the data of the Producer Price Index (PPI) of the textile industry from January 2018 to December 2023 from the official website of the National Bureau of Statistics, this paper constructs the ARIMA(1,2,1) model using the time series analysis method, and conducts an in-depth fitting analysis of the model. By using the established model to predict the PPI monthly data for the whole year of 2024 for 12 months, and comparing the prediction results with the actual values in detail, the results show that the relative error of the predicted value is far less than 5%, which fully verifies that the model has good fit and prediction accuracy. In addition, when making future predictions, it is necessary to fully consider the influence of external factors and adjust and optimize the model in a timely manner to improve the accuracy and reliability of the prediction.展开更多
本文给出了定义在有限域F2上的任意多个非线性反馈移位寄存器的互馈联结的特征函数表达式。This paper presents the characteristic function expressions for the mutual feedback connections of an arbitrary number of nonlinear f...本文给出了定义在有限域F2上的任意多个非线性反馈移位寄存器的互馈联结的特征函数表达式。This paper presents the characteristic function expressions for the mutual feedback connections of an arbitrary number of nonlinear feedback shift registers defined over the finite field F2.展开更多
由于无人机能够灵活部署,因此可以帮助提高覆盖范围和通信质量。本文考虑了一种无人机辅助的移动边缘计算系统,其中配备有计算资源的无人机可以向附近的用户设备提供卸载服务。用户将部分计算任务卸载到无人机,而其余任务在用户本地执...由于无人机能够灵活部署,因此可以帮助提高覆盖范围和通信质量。本文考虑了一种无人机辅助的移动边缘计算系统,其中配备有计算资源的无人机可以向附近的用户设备提供卸载服务。用户将部分计算任务卸载到无人机,而其余任务在用户本地执行。我们的目标是通过联合优化用户任务调度、任务卸载比率、传输功率、无人机飞行角度和飞行速度到达最小化系统成本的目的。并且考虑到该优化问题是非凸的,我们提出了一种基于深度确定性策略梯度的强化学习计算卸载算法。通过该算法,我们可以在不可控的动态环境中获得最优的计算卸载策略。并且通过仿真结果表明,该算法优于其他强化学习算法。 Due to the flexible deployment of drones, they can help improve coverage and communication quality. This paper considers a UAV assisted mobile edge computing system, in which the UAV equipped with computing resources can provide unloading services to nearby user devices. Users offload some computing tasks to the drone, while the remaining tasks are executed locally by the user. Our goal is to minimize system costs by jointly optimizing user task scheduling, task offloading ratio, transmission power, drone flight angle, and flight speed. And considering that the optimization problem is non-convex, we propose a reinforcement learning computation offloading algorithm based on Soft Actor Critic. Through this algorithm, we can obtain the optimal computation offloading strategy in uncontrollable dynamic environments. And the simulation results show that this algorithm is superior to other reinforcement learning algorithms.展开更多
潜艇作为现代海战中不可或缺的作战平台,承担着重要的战略和战术任务。其隐蔽性和高机动性使其成为海上作战中的强大力量。然而,随着鱼雷技术的不断进步,防御鱼雷攻击成为潜艇在水下作战中的一项重大挑战。本文研究了潜艇在水下作战中...潜艇作为现代海战中不可或缺的作战平台,承担着重要的战略和战术任务。其隐蔽性和高机动性使其成为海上作战中的强大力量。然而,随着鱼雷技术的不断进步,防御鱼雷攻击成为潜艇在水下作战中的一项重大挑战。本文研究了潜艇在水下作战中防御鱼雷攻击的最优策略问题,将潜艇与鱼雷的对抗转化为追逃博弈模型,假设敌方鱼雷攻击潜艇时,潜艇通过改变航向来规避攻击。通过应用博弈论中的零和博弈理论和矩阵博弈模型,本文分析了潜艇和鱼雷在有限策略集合下的最优对策。研究表明,基于潜艇和鱼雷的运动学特性及其有限策略,构建的收益矩阵能够为双方提供最优策略。最后通过具体算例与数值仿真验证了模型的合理性与可行性。Submarines constitute an indispensable asset in modern naval warfare, executing pivotal strategic and tactical missions. Their inherent stealth and superior maneuverability empower them to operate covertly in contested maritime environments. However, continuous advancements in torpedo technology have rendered the development of effective countermeasures against torpedo attacks a critical challenge in underwater combat. This paper formulates the submarine torpedo counter-measure problem as an optimal control problem within the framework of pursuit-evasion game theory. Specifically, the engagement is modeled as a zero-sum differential game where the submarine employs evasive maneuvers, principally through heading adjustments to mitigate the threat posed by an incoming torpedo. By employing a matrix game model defined over a finite discrete strategy set, we derive the Nash equilibrium solutions. The constructed payoff matrix, based on the kinematic constraints and maneuverability limitations of the submarine and torpedo, facilitates the determination of optimal strategies for both adversaries. Numerical simulations and case studies further validate the analytical robustness and practical feasibility of the proposed model in realistic underwater combat scenarios.展开更多
文章挖掘中医药学领域过去30年的研究主题,总结中医药研究主题的主流、变迁及演化,爬取中医药学领域硕博论文及权威期刊,划分时间段分析研究方向与方法,运用词云图、词频统计、LDA主题模型分析研究主题热点。查找中医药学领域的硕博论...文章挖掘中医药学领域过去30年的研究主题,总结中医药研究主题的主流、变迁及演化,爬取中医药学领域硕博论文及权威期刊,划分时间段分析研究方向与方法,运用词云图、词频统计、LDA主题模型分析研究主题热点。查找中医药学领域的硕博论文及期刊,最终整合得到14个主要研究主题。硕博论文主要研究信号通路,中药和疾病都有涉及;《中国中药杂志》以中药研究和统计分析为主;《中医杂志》更关注具体疾病的诊治。LDA主题模型能有效挖掘中医药学文献的研究主题,80%都能被相应领域的综述类文献所验证。This article explores the research topics in the field of traditional Chinese medicine over the past 30 years, summarizes the mainstream, changes, and evolution of traditional Chinese medicine research topics, crawls master’s and doctoral theses and authoritative journals in the field of traditional Chinese medicine, divides time periods to analyze research directions and methods, and uses word cloud maps, word frequency statistics, and LDA topic models to analyze research topic hotspots. Analyzing master’s and doctoral theses and journals in the field of traditional Chinese medicine, 14 main research topics were ultimately integrated. The master’s and doctoral theses mainly focus on signal pathways, including traditional Chinese medicine and diseases;China Journal of Chinese Materia Medica focuses on research and statistical analysis of traditional Chinese medicine;and Journal of Traditional Chinese Medicine focus more on the diagnosis and treatment of specific diseases. The LDA topic model can effectively explore research topics in traditional Chinese medicine literature, and 80% of them can be validated by relevant literature reviews in the field.展开更多
本文将基于模型的策略迭代方法推广到了分布式时滞系统的线性二次最优控制问题(LQR)的求解,证明了由该迭代方法得到的性能指标是递减的,且控制器收敛于最优控制器。This paper extends the model-based policy iteration method to the ...本文将基于模型的策略迭代方法推广到了分布式时滞系统的线性二次最优控制问题(LQR)的求解,证明了由该迭代方法得到的性能指标是递减的,且控制器收敛于最优控制器。This paper extends the model-based policy iteration method to the solution of the Linear Quadratic Regulator (LQR) problem for distributed delayed systems. It is demonstrated that the performance criterion obtained through this iterative method is monotonically decreasing, and the controller converges to the optimal controller.展开更多
本文在双人非合作马尔科夫博弈模型下,引入了一种策略度量指标,将保守策略推广到了双智能体情形,给出了一种保守策略梯度和策略改进的条件。这为双人非合作博弈中寻找保守策略下的纳什均衡提供了一定基础和改进方向。In this paper, a p...本文在双人非合作马尔科夫博弈模型下,引入了一种策略度量指标,将保守策略推广到了双智能体情形,给出了一种保守策略梯度和策略改进的条件。这为双人非合作博弈中寻找保守策略下的纳什均衡提供了一定基础和改进方向。In this paper, a policy metric is introduced under the two-player non-cooperative Markov game model, which generalizes the conservative policy to the two-agent case, and gives a conservative policy gradient and the conditions for policy improvement. This provides a certain foundation and improvement direction for finding Nash equilibrium under policy in two-player non-cooperative game.展开更多
本文针对传统追逃微分博弈模型在现实复杂环境下,特别是面对不完全信息和计算复杂度时求解困难的问题,创新性提出了一种基于柔性执行者–评论家(Soft Actor-Critic, SAC)算法的改进多智能体强化学习方法,应用于无人机追捕单一智能目标...本文针对传统追逃微分博弈模型在现实复杂环境下,特别是面对不完全信息和计算复杂度时求解困难的问题,创新性提出了一种基于柔性执行者–评论家(Soft Actor-Critic, SAC)算法的改进多智能体强化学习方法,应用于无人机追捕单一智能目标的微分博弈问题。SAC算法在追逃微分博弈中的优势体现在其自然实现了混合策略的概念,能够通过随机性来应对对手的动态变化,且具有较强的探索能力、稳定性和鲁棒性。与其他强化学习算法相比,SAC更适合处理不确定性强、对手行为复杂、动作空间连续的博弈问题。本文假设在部分可观测的环境下,追逐者和逃避者均无法知晓全部信息,仅能通过环境中的部分信息进行决策。为了解决这一连续优化问题,本文采用多智能体柔性执行者–评论家(multi-agent Soft Actor-Critic, MASAC)算法,使追逃双方智能体通过与环境的交互学习各自的最优策略。最终,本文通过测试展示了在部分可观测环境下,改进的多智能体强化学习方法在无人机追捕–逃避场景中的适用性与应用潜力。This paper addresses the difficulty in solving traditional pursuit-evasion differential game models in complex real-world environments, especially when dealing with incomplete information and computational complexity. An innovative solution is proposed in the form of an improved multi-agent reinforcement learning method based on the Soft Actor-Critic (SAC) algorithm, applied to the differential game problem of unmanned aerial vehicles (UAVs) pursuing a single intelligent target. The advantage of the SAC algorithm in pursuit-evasion differential games lies in its natural implementation of the mixed strategy concept, allowing it to handle dynamic changes in the opponent’s behavior through randomness, while exhibiting strong exploration capabilities, stability, and robustness. Compared to other reinforcement learning algorithms, SAC is better suited for handling games with strong uncertainty, complex opponent behaviors, and continuous action spaces. In this paper, we assume a partially observable environment where both the pursuer and evader are unaware of the full information and can only make decisions based on partial environmental observations. To address this continuous optimization problem, we adopt the multi-agent Soft Actor-Critic (MASAC) algorithm, enabling both agents in the pursuit-evasion scenario to learn their optimal strategies through interactions with the environment. Ultimately, through testing, this paper demonstrates the applicability and potential of the improved multi-agent reinforcement learning method in UAV pursuit-evasion scenarios within partially observable environments.展开更多
文摘该研究提出了一种基于特征提取的深度神经网络(ResNet),用于对局灶性肝脏病变的超声波图像进行分类。这些优势使其能够有效地从超声图像中提取肝脏损伤的迹象,并实现准确的分类。基于包含血管瘤、脂肪肝、肝转移瘤、肝囊肿和正常肝脏的超声图像数据集,该模型在实验验证中提供了出色的分类效果。在测试集上,该模型的准确率达到了93.99%。本文将该模型与AlexNet和VGGNet模型进行了比较,结果表明作者获得的模型分类效果更好,在准确率、参数数量和学习效率方面都有显著优势,并且具有很强的泛化能力。这项研究在病灶性肝脏病变的超声波图像分类任务中具有潜在的应用价值,可为临床医生提供准确、快速的辅助诊断工具。This study proposes a feature extraction-based deep neural network (ResNet) for classifying ultrasound images of focal liver lesions. These advantages enable it to effectively extract signs of liver damage from ultrasound images and achieve accurate classification. Based on an ultrasound image dataset containing hemangioma, fatty liver, liver metastases, liver cysts, and normal liver, the model provided excellent classification results in experimental validation. On the test set, the model achieved an accuracy of 93.99%. This article compares this model with AlexNet and VGGNet models. The results show that the model obtained by the author has better classification results, has significant advantages in accuracy, number of parameters, and learning efficiency, and has strong generalization ability. This study has potential application value in the task of ultrasonic image classification of focal liver lesions and can provide clinicians with an accurate and rapid auxiliary diagnostic tool.
文摘本文采用自适应控制和事件触发策略研究了矩阵加权网络下多智能体系统的一致性问题。不同于传统的纯量加权网络,智能体及其邻居之间的通信由正定或半正定的矩阵刻画。利用矩阵权值将更有利于反映智能体之间的逻辑依赖关系。结合动态事件触发策略设计了两种不同的自适应控制协议,即基于边的自适应协议和基于节点的自适应协议,这些控制协议具有灵活调整控制输入的性能。给出了实现一致性的充分条件,并分析了芝诺行为的排除。最后,通过数值仿真验证了理论分析的有效性。This paper endeavors to investigate the consensus problem for multi-agent systems under matrix-weighted networks by employing adaptive control and event-triggered strategies. Different from conventional scalar-weighted networks, the interconnections among agents and their neighbors are characterized by positive definite or positive semi-definite matrices. The utilization of matrix weights is more beneficial to reflect the logical inter-dependency among agents. Two different adaptive control protocols, namely edge-based adaptive protocol and node-based adaptive protocol, are designed in combination with dynamic event-triggered strategies. These protocols can facilitate the flexible manipulation of control inputs. Sufficient conditions for achieving consensus are provided and the exclusion of Zeno behavior is analyzed. Finally, the validity of the theoretical analysis is verified through numerical simulation.
文摘乳腺癌是女性常见恶性肿瘤之一,严重威胁女性健康。在乳腺癌手术中,准确测量乳房和切除组织的体积对于手术规划和乳腺重建至关重要。本研究构建了一个基于深度学习的分割框架,用于乳房和切除组织的MRI体积分割。该框架不仅展现了良好的分割精度,还涵盖了偏置场矫正这一关键预处理步骤。我们收集并手动标注了一个包含47例患者的MRI数据集,这些数据涵盖了不同的年龄、乳房大小和四类采集参数。在交叉验证中,U-Net网络在全乳分割和切除组织分割任务中表现最佳,平均Dice系数分别为96.54和92.37。在测试集上,U-Net网络同样展现了优异的分割效果,平均Dice系数分别为94.23和84.53。实验结果表明,所提出的框架能够精准且高效地量化全乳体积和切除组织体积,为临床乳腺手术提供数据支持。Breast cancer is one of the most common malignant tumors in women, which seriously threatens women’s health. In breast cancer surgery, accurate measurement of breast and excised tissue volume is critical for surgical planning and breast reconstruction. This study developed a deep learning based segmentation framework for MRI volume segmentation of breasts and excised tissues. This framework not only demonstrates good segmentation accuracy, but also covers the key preprocessing step of bias field correction. We collected and manually annotated an MRI dataset containing 47 patients, covering different ages, breast sizes, and four types of acquisition parameters. In cross validation, the U-Net network performed the best in both whole milk segmentation and excised tissue segmentation tasks, with average Dice coefficients of 96.54 and 92.37, respectively. On the test set, the U-Net network also demonstrated excellent segmentation performance, with average Dice coefficients of 94.23 and 84.53, respectively. The experimental results indicate that the proposed framework can accurately and efficiently quantify total breast volume and excised tissue volume, providing data support for clinical breast surgery.
文摘本文主要研究在月光型顶点算子代数中满足一定条件的2对Ising向量生成的顶点算子代数的结构,这2对Ising向量分别生成1个3A代数,并且生成的2个3A代数的交包含一个同构于L(4/5, 0)⊕L(4/5, 3)的子顶点算子代数,本文证明了其一共有3种可能的顶点算子代数结构。In this paper, we mainly study the vertex operator algebra generated by two pairs of Ising vectors in the moonshine type vertex operator algebra. These two pairs of Ising vectors each generate one 3A algebra, and the intersection of the two generated 3A algebras contains a subvertex operator subalgebra that is isomorphic to L(4/5, 0)⊕L(4/5, 3). We have shown that there are three possible structures of vertex operators algebraic.
文摘本文基于国家统计局官网关于纺织行业2018年01月~2023年12月的工业生产者出厂价格指数(PPI)数据,运用时间序列分析方法构建了ARIMA(1,2,1)模型,并对该模型进行了深入的拟合分析。利用所建立的模型对2024年全年12个月的PPI月度数据进行预测,并将预测结果与实际值进行对比,结果显示预测值的相对误差远低于5%,充分验证了模型具有良好的拟合度和预测精度。此外,在进行未来预测时,需要充分考虑外部因素的影响,并对模型进行适时的调整和优化,以提高预测的准确性和可靠性。Based on the data of the Producer Price Index (PPI) of the textile industry from January 2018 to December 2023 from the official website of the National Bureau of Statistics, this paper constructs the ARIMA(1,2,1) model using the time series analysis method, and conducts an in-depth fitting analysis of the model. By using the established model to predict the PPI monthly data for the whole year of 2024 for 12 months, and comparing the prediction results with the actual values in detail, the results show that the relative error of the predicted value is far less than 5%, which fully verifies that the model has good fit and prediction accuracy. In addition, when making future predictions, it is necessary to fully consider the influence of external factors and adjust and optimize the model in a timely manner to improve the accuracy and reliability of the prediction.
文摘本文给出了定义在有限域F2上的任意多个非线性反馈移位寄存器的互馈联结的特征函数表达式。This paper presents the characteristic function expressions for the mutual feedback connections of an arbitrary number of nonlinear feedback shift registers defined over the finite field F2.
文摘由于无人机能够灵活部署,因此可以帮助提高覆盖范围和通信质量。本文考虑了一种无人机辅助的移动边缘计算系统,其中配备有计算资源的无人机可以向附近的用户设备提供卸载服务。用户将部分计算任务卸载到无人机,而其余任务在用户本地执行。我们的目标是通过联合优化用户任务调度、任务卸载比率、传输功率、无人机飞行角度和飞行速度到达最小化系统成本的目的。并且考虑到该优化问题是非凸的,我们提出了一种基于深度确定性策略梯度的强化学习计算卸载算法。通过该算法,我们可以在不可控的动态环境中获得最优的计算卸载策略。并且通过仿真结果表明,该算法优于其他强化学习算法。 Due to the flexible deployment of drones, they can help improve coverage and communication quality. This paper considers a UAV assisted mobile edge computing system, in which the UAV equipped with computing resources can provide unloading services to nearby user devices. Users offload some computing tasks to the drone, while the remaining tasks are executed locally by the user. Our goal is to minimize system costs by jointly optimizing user task scheduling, task offloading ratio, transmission power, drone flight angle, and flight speed. And considering that the optimization problem is non-convex, we propose a reinforcement learning computation offloading algorithm based on Soft Actor Critic. Through this algorithm, we can obtain the optimal computation offloading strategy in uncontrollable dynamic environments. And the simulation results show that this algorithm is superior to other reinforcement learning algorithms.
文摘潜艇作为现代海战中不可或缺的作战平台,承担着重要的战略和战术任务。其隐蔽性和高机动性使其成为海上作战中的强大力量。然而,随着鱼雷技术的不断进步,防御鱼雷攻击成为潜艇在水下作战中的一项重大挑战。本文研究了潜艇在水下作战中防御鱼雷攻击的最优策略问题,将潜艇与鱼雷的对抗转化为追逃博弈模型,假设敌方鱼雷攻击潜艇时,潜艇通过改变航向来规避攻击。通过应用博弈论中的零和博弈理论和矩阵博弈模型,本文分析了潜艇和鱼雷在有限策略集合下的最优对策。研究表明,基于潜艇和鱼雷的运动学特性及其有限策略,构建的收益矩阵能够为双方提供最优策略。最后通过具体算例与数值仿真验证了模型的合理性与可行性。Submarines constitute an indispensable asset in modern naval warfare, executing pivotal strategic and tactical missions. Their inherent stealth and superior maneuverability empower them to operate covertly in contested maritime environments. However, continuous advancements in torpedo technology have rendered the development of effective countermeasures against torpedo attacks a critical challenge in underwater combat. This paper formulates the submarine torpedo counter-measure problem as an optimal control problem within the framework of pursuit-evasion game theory. Specifically, the engagement is modeled as a zero-sum differential game where the submarine employs evasive maneuvers, principally through heading adjustments to mitigate the threat posed by an incoming torpedo. By employing a matrix game model defined over a finite discrete strategy set, we derive the Nash equilibrium solutions. The constructed payoff matrix, based on the kinematic constraints and maneuverability limitations of the submarine and torpedo, facilitates the determination of optimal strategies for both adversaries. Numerical simulations and case studies further validate the analytical robustness and practical feasibility of the proposed model in realistic underwater combat scenarios.
文摘文章挖掘中医药学领域过去30年的研究主题,总结中医药研究主题的主流、变迁及演化,爬取中医药学领域硕博论文及权威期刊,划分时间段分析研究方向与方法,运用词云图、词频统计、LDA主题模型分析研究主题热点。查找中医药学领域的硕博论文及期刊,最终整合得到14个主要研究主题。硕博论文主要研究信号通路,中药和疾病都有涉及;《中国中药杂志》以中药研究和统计分析为主;《中医杂志》更关注具体疾病的诊治。LDA主题模型能有效挖掘中医药学文献的研究主题,80%都能被相应领域的综述类文献所验证。This article explores the research topics in the field of traditional Chinese medicine over the past 30 years, summarizes the mainstream, changes, and evolution of traditional Chinese medicine research topics, crawls master’s and doctoral theses and authoritative journals in the field of traditional Chinese medicine, divides time periods to analyze research directions and methods, and uses word cloud maps, word frequency statistics, and LDA topic models to analyze research topic hotspots. Analyzing master’s and doctoral theses and journals in the field of traditional Chinese medicine, 14 main research topics were ultimately integrated. The master’s and doctoral theses mainly focus on signal pathways, including traditional Chinese medicine and diseases;China Journal of Chinese Materia Medica focuses on research and statistical analysis of traditional Chinese medicine;and Journal of Traditional Chinese Medicine focus more on the diagnosis and treatment of specific diseases. The LDA topic model can effectively explore research topics in traditional Chinese medicine literature, and 80% of them can be validated by relevant literature reviews in the field.
文摘本文将基于模型的策略迭代方法推广到了分布式时滞系统的线性二次最优控制问题(LQR)的求解,证明了由该迭代方法得到的性能指标是递减的,且控制器收敛于最优控制器。This paper extends the model-based policy iteration method to the solution of the Linear Quadratic Regulator (LQR) problem for distributed delayed systems. It is demonstrated that the performance criterion obtained through this iterative method is monotonically decreasing, and the controller converges to the optimal controller.
文摘本文在双人非合作马尔科夫博弈模型下,引入了一种策略度量指标,将保守策略推广到了双智能体情形,给出了一种保守策略梯度和策略改进的条件。这为双人非合作博弈中寻找保守策略下的纳什均衡提供了一定基础和改进方向。In this paper, a policy metric is introduced under the two-player non-cooperative Markov game model, which generalizes the conservative policy to the two-agent case, and gives a conservative policy gradient and the conditions for policy improvement. This provides a certain foundation and improvement direction for finding Nash equilibrium under policy in two-player non-cooperative game.
文摘本文针对传统追逃微分博弈模型在现实复杂环境下,特别是面对不完全信息和计算复杂度时求解困难的问题,创新性提出了一种基于柔性执行者–评论家(Soft Actor-Critic, SAC)算法的改进多智能体强化学习方法,应用于无人机追捕单一智能目标的微分博弈问题。SAC算法在追逃微分博弈中的优势体现在其自然实现了混合策略的概念,能够通过随机性来应对对手的动态变化,且具有较强的探索能力、稳定性和鲁棒性。与其他强化学习算法相比,SAC更适合处理不确定性强、对手行为复杂、动作空间连续的博弈问题。本文假设在部分可观测的环境下,追逐者和逃避者均无法知晓全部信息,仅能通过环境中的部分信息进行决策。为了解决这一连续优化问题,本文采用多智能体柔性执行者–评论家(multi-agent Soft Actor-Critic, MASAC)算法,使追逃双方智能体通过与环境的交互学习各自的最优策略。最终,本文通过测试展示了在部分可观测环境下,改进的多智能体强化学习方法在无人机追捕–逃避场景中的适用性与应用潜力。This paper addresses the difficulty in solving traditional pursuit-evasion differential game models in complex real-world environments, especially when dealing with incomplete information and computational complexity. An innovative solution is proposed in the form of an improved multi-agent reinforcement learning method based on the Soft Actor-Critic (SAC) algorithm, applied to the differential game problem of unmanned aerial vehicles (UAVs) pursuing a single intelligent target. The advantage of the SAC algorithm in pursuit-evasion differential games lies in its natural implementation of the mixed strategy concept, allowing it to handle dynamic changes in the opponent’s behavior through randomness, while exhibiting strong exploration capabilities, stability, and robustness. Compared to other reinforcement learning algorithms, SAC is better suited for handling games with strong uncertainty, complex opponent behaviors, and continuous action spaces. In this paper, we assume a partially observable environment where both the pursuer and evader are unaware of the full information and can only make decisions based on partial environmental observations. To address this continuous optimization problem, we adopt the multi-agent Soft Actor-Critic (MASAC) algorithm, enabling both agents in the pursuit-evasion scenario to learn their optimal strategies through interactions with the environment. Ultimately, through testing, this paper demonstrates the applicability and potential of the improved multi-agent reinforcement learning method in UAV pursuit-evasion scenarios within partially observable environments.
