Using Hybrid Deep Learning Model to Detect Anomaly Traffic Efficiently
Yi-Jen Su, Tso-Lin Kuo, Shan-Hsiung Hsieh, Chao-Ho Chen, Tsong-Yi Chen, Li-Chin Wu · 2022 IEEE 11th Global Conference on Consumer Electronics (GCCE) · 2022
Internet traffic has escalated throughout recent years, resulting in information security evolving into a severe issue. Therefore, utilizing Intrusion Detection System to monitor malicious attacks in achieving high network protection has become a significant research issue. The study proposes a novel IDS (Intrusion Detection System) to detect network malicious behaviors with a hybrid CNN+LSTM model. To effectively reduce the classification time, PCA is adopted for dimensional reduction from 47 attributes to 31 components, covering 99% of feature information in the UNSW-NB15 dataset. The accuracy of our proposed solution reaches 98.24% along with a much shorter detection time.