
IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 93235 - 93260
Published: Jan. 1, 2024
Cyber Threat Detection (CTD) is subject to complicated and rapidly accelerating developments. Poor accuracy, high learning complexity, limited scalability, a false positive rate are problems that CTD encounters. Deep Learning defense mechanisms aim build effective models for threat detection protection allowing them adapt the complex ever-accelerating changes in field of CTD. Furthermore, swarm intelligence algorithms have been developed tackle optimization challenges. In this paper, Chaotic Zebra Optimization Long-Short Term Memory (CZOLSTM) algorithm proposed. The proposed hybrid between Algorithm (CZOA) feature selection LSTM cyber classification CSE-CIC-IDS2018 dataset. Invoking chaotic map CZOLSTM can improve diversity search avoid trapping local minimum. evaluating effectiveness newly CZOLSTM, binary multi-class classifications considered. acquired outcomes demonstrate efficiency implemented improvements across many other algorithms. When comparing performance detection, it outperforms six innovative deep five classification. Other evaluation criteria such as recall, F1 score, precision also used comparison. results showed best accuracy was achieved using 99.83%, with F1-score 99.82%, recall 99.82%. among compared
Language: Английский