Evidenctial classifier based on deep learning and D-S theory for network intrusion detection. Using CNN, RNN and LSTM as feature extractors. Experiments on datasets NSL-KDD and UNSW-NB15 datasets demonstrate that evidenctial classifier is able to make cautious decisions by assigning confusing samples to multi-class sets while maintaining the accuracy.
The required libraries and their version:
python == 3.7.10
tensorflow == 2.4.1