The rapid development of network technology has brought convenience to people’s daily life,and the Internet user group has continued to grow.However,the Internet is also facing new challenges at the same time.Cyber-attacks are increasingly developing in the direction of deep penetration,long-term stealing,and strategic control,which have brought severe challenges to social network information systems.Therefore,it has great important research significance to protect the security of information systems and improve the ability to detect and prevent abnormal behaviors of the network.At present,the theory of artificial intelligence has developed rapidly and has the advantages of adapting to changes of new data and high accuracy.It plays an important role in computer vision,speech recognition and other fields.In this paper,artificial intelligence algorithm is used to study the abnormal behavior in the network.This paper proposes three research scenarios for detecting network abnormal behavior which are based on lifting algorithm to identify cyber-attacks and disturbances in smart grids,analyzing network router logs to detect abnormal network behavior based on neural network,and based on semantic and spatial features to analyze malicious URLs.In the above scenarios,the common goal is to process data and build models to further mine hidden abnormal network behaviors.The main contents are as follows:In research point 1,feature engineering is performed on the log data of the smart grid,16 new features are constructed based on the actual physical meaning.At the same time,a lifting algorithm model based on random forest is proposed for the abnormal behavior detection of smart grid.Experimental results show that the proposed method can better identify abnormal behaviors such as faults and network attacks.In research point 2,an abnormal network behavior detection model based on neural network is adopted.This model is used to detect and analyze the router board log,and then mine abnormal behavior in the log.Apart from that,a semantic forest and a feature library is constructed according to the attribute information and different behavior types.Experimental results on datasets show that it can effectively detect abnormal situations such as log surge caused by router faults or network attacks.In research point 3,this paper proposes a detection model framework based on semantic and spatial features for uniform resource locator data such as web pages or links,and combines with convolutional neural network model.Experimental results show that the model performs better than other traditional machine learning algorithms when dealing with the imbalance of positive and negative data samples. |