A bidirectional LSTM deep learning approach for intrusion detection

作者：

Highlights：

• The problem of high rates of raising false alarms in IDSs is considered.

• It is difficult for some existing methods to detect U2R and R2L attacks.

• To tackle these challenges, a deep learning-based BiDLSTM model is developed.

• The proposed method exhibits better accuracies than the conventional LSTM.

• The proposed BiDLSTM also outperformed many state-of-the-art methods.

摘要

•The problem of high rates of raising false alarms in IDSs is considered.•It is difficult for some existing methods to detect U2R and R2L attacks.•To tackle these challenges, a deep learning-based BiDLSTM model is developed.•The proposed method exhibits better accuracies than the conventional LSTM.•The proposed BiDLSTM also outperformed many state-of-the-art methods.

论文关键词：Machine learning,Deep learning,Recurrent neural networks,Bidirectional LSTM,Intrusion detection

论文评审过程：Received 19 June 2020, Revised 27 June 2021, Accepted 29 June 2021, Available online 16 July 2021, Version of Record 20 July 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115524