| This paper takes the security of medical information system as the research subject,and designs an intrusion detection system based on deep learning(DL).This system makes full use of the advantage of unsupervised learning in deep learning,and detects intrusion behavior accurately by the fusion algorithm.This research trys to deal with network intrusion behaviors by fusing networks of convolutional neural to networks of long-term and short-term memory recurrent neural.It focuses on detection performance which is effected by different input dimension,different convolution kernel sizes and LSTM memory modules.The research shows that the fused network can detect network intrusion behaviors better,which provides a new feasible method for detecting network intrusion.This paper first introduces the value and significance of this research and the research situation of this subject at home and abroad;discusses the relevant concepts,and specifically analyzes the key points of the development of intrusion detection technology,firewall concept,firewall classification,convolution neural network structure and other aspects;then,the construction of medical information system in a medical institution is explored,and the hospital is analyzed in detail The existing intrusion detection methods,system security requirements and system security management objectives;then,in view of the defects in the existing intrusion detection system,an optimal design is carried out.A network intrusion detection method with convolutional neural network as the core is designed,and in-depth research is carried out from data collection,Bi LSTM model design and other aspects.Finally,the simulation test shows that under the specific conditions of input dimension and convolution kernel value,the accuracy and false alarm rate are improved,which proves that the optimization design does improve the accuracy of detecting attack behavior.It provides a new idea for the future development of network intrusion detection. |
PDF Full Text Request