Enhanced Cybersecurity via Deep LSTM Networks for Intrusion Detection

DOI

No DOI

Published In

Volume 3, Issue 3 (2024)

Published Online

September-December 2024

Pages

17-19

Article Type

Research Article

Check for updates

No DOI available for cross-reference

Deep Long Short-Term Memory Recurrent Neural network (LSTM-RNN) methodology is consists of pre-processing, training and testing phases. The raw data attributes consist of numerical and non-numerical values. Non-numerical values need to be conversion of numerical values because the LSTM-RNN model requires numerical attribute values as input. The numericalization process can be done with one-hot encoding. One-hot encoding assigns unique feature values to the non-numerical features. Some numerical zed data attributes consist of large feature value and some attributes consist of minimum value. The difference between minimum feature value and maximum feature value is very large. This difference affects the original feature values. The normalization process avoids the effectiveness of the original feature values. Normalization could be done with min-max normalization.