Internet services and web-based applications play pivotal roles in various sensitive domains, encompassing e-commerce, e-learning, e-healthcare, and e-payment. However, safeguarding these services poses a significant ...Internet services and web-based applications play pivotal roles in various sensitive domains, encompassing e-commerce, e-learning, e-healthcare, and e-payment. However, safeguarding these services poses a significant challenge, as the need for robust security measures becomes increasingly imperative. This paper presented an innovative method based on differential analyses to detect abrupt changes in network traffic characteristics. The core concept revolves around identifying abrupt alterations in certain characteristics such as input/output volume, the number of TCP connections, or DNS queries—within the analyzed traffic. Initially, the traffic is segmented into distinct sequences of slices, followed by quantifying specific characteristics for each slice. Subsequently, the distance between successive values of these measured characteristics is computed and clustered to detect sudden changes. To accomplish its objectives, the approach combined several techniques, including propositional logic, distance metrics (e.g., Kullback-Leibler Divergence), and clustering algorithms (e.g., K-means). When applied to two distinct datasets, the proposed approach demonstrates exceptional performance, achieving detection rates of up to 100%.展开更多
树增强朴素贝叶斯(TAN)分类器在模型的复杂性和分类精度之间实现较好折衷,成为当前分类器学习的一个研究热点.为了提高 TAN 分类器的分类准确率,本文提出一种基于 KL 距离的 TAN 分类器判别性学习方法.首先用 EAR 方法学习 TAN 分类器...树增强朴素贝叶斯(TAN)分类器在模型的复杂性和分类精度之间实现较好折衷,成为当前分类器学习的一个研究热点.为了提高 TAN 分类器的分类准确率,本文提出一种基于 KL 距离的 TAN 分类器判别性学习方法.首先用 EAR 方法学习 TAN 分类器的结构,然后用基于 KL 距离的目标函数优化 TAN 的参数.在标准数据集上的实验结果表明,用该方法学习的 TAN 分类器具有较高的分类精度.展开更多
Currently,a surge in the number of spacecraft and fragments is observed,leading to more frequent breakup events in low Earth orbits(LEOs).The causes of these events are being identified,and specific triggers,such as c...Currently,a surge in the number of spacecraft and fragments is observed,leading to more frequent breakup events in low Earth orbits(LEOs).The causes of these events are being identified,and specific triggers,such as collisions or explosions,are being examined for their importance to space traffic management.Backward propagation methods were employed to trace the origins of these types of breakup events.Simulations were conducted using the NASA standard breakup model,and satellite Hitomi’s breakup was analyzed.Kullback-Leibler(KL)divergences,Euclidean 2-norms,and Jensen-Shannon(JS)divergences were computed to deduce potential types of breakups and the associated fragmentation masses.In the simulated case,a discrepancy of 22.12 s between the estimated and actual time was noted.Additionally,the breakup of the Hitomi satellite was estimated to have occurred around UTC 1:49:26.4 on March 26,2016.This contrasts with the epoch provided by the Joint Space Operation Center,which was estimated to be at 1:42 UTC±11 min.From the findings,it was suggested that the techniques introduced in the study can be effectively used to trace the origins of short-term breakup events and to deduce the types of collisions and fragmentation masses under certain conditions.展开更多
文摘Internet services and web-based applications play pivotal roles in various sensitive domains, encompassing e-commerce, e-learning, e-healthcare, and e-payment. However, safeguarding these services poses a significant challenge, as the need for robust security measures becomes increasingly imperative. This paper presented an innovative method based on differential analyses to detect abrupt changes in network traffic characteristics. The core concept revolves around identifying abrupt alterations in certain characteristics such as input/output volume, the number of TCP connections, or DNS queries—within the analyzed traffic. Initially, the traffic is segmented into distinct sequences of slices, followed by quantifying specific characteristics for each slice. Subsequently, the distance between successive values of these measured characteristics is computed and clustered to detect sudden changes. To accomplish its objectives, the approach combined several techniques, including propositional logic, distance metrics (e.g., Kullback-Leibler Divergence), and clustering algorithms (e.g., K-means). When applied to two distinct datasets, the proposed approach demonstrates exceptional performance, achieving detection rates of up to 100%.
文摘树增强朴素贝叶斯(TAN)分类器在模型的复杂性和分类精度之间实现较好折衷,成为当前分类器学习的一个研究热点.为了提高 TAN 分类器的分类准确率,本文提出一种基于 KL 距离的 TAN 分类器判别性学习方法.首先用 EAR 方法学习 TAN 分类器的结构,然后用基于 KL 距离的目标函数优化 TAN 的参数.在标准数据集上的实验结果表明,用该方法学习的 TAN 分类器具有较高的分类精度.
基金Supported by the National Natural Science Foundation of China(61371170)Funding of Jiangsu Innovation Program for Graduate Education(CXLX13_154)+1 种基金the Fundamental Research Funds for the Central Universities,the Priority Academic Program Development of Jiangsu Higher Education Institutions(PADA)Key Laboratory of Radar Imaging and Microwave Photonics,Ministry of Education,Nanjing University of Aeronautics and Astronautics
基金grateful to the National Key R&D Program of China(Grant No.2022ZD0117301)for funding this study。
文摘Currently,a surge in the number of spacecraft and fragments is observed,leading to more frequent breakup events in low Earth orbits(LEOs).The causes of these events are being identified,and specific triggers,such as collisions or explosions,are being examined for their importance to space traffic management.Backward propagation methods were employed to trace the origins of these types of breakup events.Simulations were conducted using the NASA standard breakup model,and satellite Hitomi’s breakup was analyzed.Kullback-Leibler(KL)divergences,Euclidean 2-norms,and Jensen-Shannon(JS)divergences were computed to deduce potential types of breakups and the associated fragmentation masses.In the simulated case,a discrepancy of 22.12 s between the estimated and actual time was noted.Additionally,the breakup of the Hitomi satellite was estimated to have occurred around UTC 1:49:26.4 on March 26,2016.This contrasts with the epoch provided by the Joint Space Operation Center,which was estimated to be at 1:42 UTC±11 min.From the findings,it was suggested that the techniques introduced in the study can be effectively used to trace the origins of short-term breakup events and to deduce the types of collisions and fragmentation masses under certain conditions.
