ENHANCING PACKET INSPECTION ACCURACY TO IDENTIFY NETWORK LAYER ATTACKS USING MACHINE LEARNING

Intrusion discovery can identify unknown attacks from network traffics and has been an effective means of network security. currently, being styles for network anomaly discovery are generally grounded on traditional machine literacy models, similar as KNN, SVM, etc. Although these styles can gain some outstanding features, they get a low delicacy, cannot handle large data, low performance and calculate heavily on homemade design of business features, which has been obsolete in the age of big data. To break the problems of low delicacy and point engineering in intrusion discovery, a business anomaly discovery model club is proposed. The club model combines BLSTM (Bidirectional Long Short-term memory) and attention mechanism. This model has got a aggregate of 5 layers. Attention mechanism is used to screen the network inflow, vector composed of packet vectors generated by the BLSTM model, which can gain the crucial features for network business bracket. In addition, we borrow multiple convolutional layers to capture the original features of business data. As multiple convolutional layers are used to reuse data samples, we relate BAT model as BAT- MC. The SoftMax classifier is used for network traffic classification. The proposed end- to- end model does not use any feature engineering skills and can automatically learn the key features of the hierarchy. It can well describe the network business geste and ameliorate the capability of anomaly discovery effectively and suitable to handle large data and good performance. We test our model on a public standard dataset, and the experimental results demonstrate our model has better performance than other comparison styles.