Based on the perspective of big data,the growth characteristics of marine science and technology talents were analyzed,and the growth of marine science and technology talents was divided into five periods:study period...Based on the perspective of big data,the growth characteristics of marine science and technology talents were analyzed,and the growth of marine science and technology talents was divided into five periods:study period,adaptation period,growth period,promotion period and stability period.Moreover,some suggestions for the training of marine science and technology talents were proposed from the aspects of students,families,schools and society.展开更多
In this paper,we study the large-time behavior of periodic solutions for parabolic conservation laws.There is no smallness assumption on the initial data.We firstly get the local existence of the solution by the itera...In this paper,we study the large-time behavior of periodic solutions for parabolic conservation laws.There is no smallness assumption on the initial data.We firstly get the local existence of the solution by the iterative scheme,then we get the exponential decay estimates for the solution by energy method and maximum principle,and obtain the global solution in the same time.展开更多
Advances in technology require upgrades in the law. One such area involves data brokers, which have thus far gone unregulated. Data brokers use artificial intelligence to aggregate information into data profiles about...Advances in technology require upgrades in the law. One such area involves data brokers, which have thus far gone unregulated. Data brokers use artificial intelligence to aggregate information into data profiles about individual Americans derived from consumer use of the internet and connected devices. Data profiles are then sold for profit. Government investigators use a legal loophole to purchase this data instead of obtaining a search warrant, which the Fourth Amendment would otherwise require. Consumers have lacked a reasonable means to fight or correct the information data brokers collect. Americans may not even be aware of the risks of data aggregation, which upends the test of reasonable expectations used in a search warrant analysis. Data aggregation should be controlled and regulated, which is the direction some privacy laws take. Legislatures must step forward to safeguard against shadowy data-profiling practices, whether abroad or at home. In the meantime, courts can modify their search warrant analysis by including data privacy principles.展开更多
基金Supported by Foundation for Humanities and Social Sciences Research Planning of Ministry of Education of Shanghai City(19YJA630058)
文摘Based on the perspective of big data,the growth characteristics of marine science and technology talents were analyzed,and the growth of marine science and technology talents was divided into five periods:study period,adaptation period,growth period,promotion period and stability period.Moreover,some suggestions for the training of marine science and technology talents were proposed from the aspects of students,families,schools and society.
基金Foundation item： Supported by the National Science Foundation of China（1107116）
文摘In this paper,we study the large-time behavior of periodic solutions for parabolic conservation laws.There is no smallness assumption on the initial data.We firstly get the local existence of the solution by the iterative scheme,then we get the exponential decay estimates for the solution by energy method and maximum principle,and obtain the global solution in the same time.
文摘Advances in technology require upgrades in the law. One such area involves data brokers, which have thus far gone unregulated. Data brokers use artificial intelligence to aggregate information into data profiles about individual Americans derived from consumer use of the internet and connected devices. Data profiles are then sold for profit. Government investigators use a legal loophole to purchase this data instead of obtaining a search warrant, which the Fourth Amendment would otherwise require. Consumers have lacked a reasonable means to fight or correct the information data brokers collect. Americans may not even be aware of the risks of data aggregation, which upends the test of reasonable expectations used in a search warrant analysis. Data aggregation should be controlled and regulated, which is the direction some privacy laws take. Legislatures must step forward to safeguard against shadowy data-profiling practices, whether abroad or at home. In the meantime, courts can modify their search warrant analysis by including data privacy principles.
文摘数据增广是提升深度学习模型性能的有效方法之一。针对多类别目标检测任务中检测性能不平衡问题,提出一种针对“短板类别”(检测性能远低于模型平均检测性能的类别)的离线数据增广方法。受Cannikin’s Law的启发,采用基于复制粘贴(copy-paste)机制的场景多样性增广方法。随机采集训练集中“短板类别”实例区域,通过相似性度量机制选取训练集中增广目标样本进行随机粘贴。为了降低随机粘贴导致的遮挡问题,采用基于自遮挡(cut-replace)机制的增广方法提升模型遮挡表达能力。通过截取样本自身区域,对特征表达最显著区域进行遮挡。实验表明,FCOS目标检测框架在PASCAL VOC数据上的平均检测精度(mean average precision,mAP)从79.10%提升到83.90%,其中短板类别更为显著,提升了20.8个百分点。在MS-COCO数据上平均检测精度提升了0.9个百分点。
