为进一步提升应用层DDoS攻击检测准确率,提出一种将流量与用户行为特征相结合且模型参数可高效更新的应用层DDoS攻击检测模型.为统一处理流量与用户行为特征的异源数据,利用多模态深度(Multimodal Deep Learning,MDL)神经网络从数据流...为进一步提升应用层DDoS攻击检测准确率,提出一种将流量与用户行为特征相结合且模型参数可高效更新的应用层DDoS攻击检测模型.为统一处理流量与用户行为特征的异源数据,利用多模态深度(Multimodal Deep Learning,MDL)神经网络从数据流量与网页日志中提取流量与用户行为深层特征后输入汇聚深度神经网络进行检测.为减少MDL神经网络参数更新时的灾难性遗忘现象,在模型参数更新过程中基于弹性权重保持(Elastic Weight Consolidation,EWC)算法为重要模型参数增加惩罚项,保持对初始训练数据集检测准确率的同时,提升对新数据集的检测性能.最后,基于K-Means算法获得模型初始训练数据集聚类,并筛选出新数据集中聚类外数据进行模型参数更新,防止EWC算法因数据相关性过高而失效.实验表明,所提应用层DDoS检测模型检测准确率可达98.2%,且相对MLP_Whole方法模型参数更新性能较好.展开更多
As the most frequent word sequence in register,lexical bundles are the basic components of discourse.Their use is of great help to the fluency and accuracy of academic writing.In the past two decades,the study of lexi...As the most frequent word sequence in register,lexical bundles are the basic components of discourse.Their use is of great help to the fluency and accuracy of academic writing.In the past two decades,the study of lexical bundles has focused on the body part of academic papers,but less on the abstract.In order to identify and compare the differences in the use of lexical bundles in English abstracts between international and Chinese authors,this paper selects English abstracts from Chinese and International high-level academic journals in the discipline of linguistics from 2018 to 2020 to build two corpora of English abstracts of academic writings:Corpus of English Abstracts from International Linguistics Journal Articles(ILJA)and Corpus of English Abstracts from Chinese Linguistics Journal Articles(CLJA).This paper analyzes the differences of English abstract writing between Chinese and international high-level linguistics journals,and provides enlightenment for abstract writing and teaching.展开更多
文摘为进一步提升应用层DDoS攻击检测准确率,提出一种将流量与用户行为特征相结合且模型参数可高效更新的应用层DDoS攻击检测模型.为统一处理流量与用户行为特征的异源数据,利用多模态深度(Multimodal Deep Learning,MDL)神经网络从数据流量与网页日志中提取流量与用户行为深层特征后输入汇聚深度神经网络进行检测.为减少MDL神经网络参数更新时的灾难性遗忘现象,在模型参数更新过程中基于弹性权重保持(Elastic Weight Consolidation,EWC)算法为重要模型参数增加惩罚项,保持对初始训练数据集检测准确率的同时,提升对新数据集的检测性能.最后,基于K-Means算法获得模型初始训练数据集聚类,并筛选出新数据集中聚类外数据进行模型参数更新,防止EWC算法因数据相关性过高而失效.实验表明,所提应用层DDoS检测模型检测准确率可达98.2%,且相对MLP_Whole方法模型参数更新性能较好.
文摘As the most frequent word sequence in register,lexical bundles are the basic components of discourse.Their use is of great help to the fluency and accuracy of academic writing.In the past two decades,the study of lexical bundles has focused on the body part of academic papers,but less on the abstract.In order to identify and compare the differences in the use of lexical bundles in English abstracts between international and Chinese authors,this paper selects English abstracts from Chinese and International high-level academic journals in the discipline of linguistics from 2018 to 2020 to build two corpora of English abstracts of academic writings:Corpus of English Abstracts from International Linguistics Journal Articles(ILJA)and Corpus of English Abstracts from Chinese Linguistics Journal Articles(CLJA).This paper analyzes the differences of English abstract writing between Chinese and international high-level linguistics journals,and provides enlightenment for abstract writing and teaching.
