Skip to main content

Table 5 Summary of deep learning model techniques

From: A critical review of intrusion detection systems in the internet of things: techniques, deployment strategy, validation strategy, attacks, public datasets and challenges

Model Learning Model Input Data Characteristics
FCNN Supervised Image, sound, etc. - No special assumptions needed to be made about the input.
-Requires a huge number of connections and network parameters.
RNN Supervised Serial, time-series -Processes sequences of data through internal data.
-Useful in IDS with time-dependent
GAN Semi-supervised various -The GAN sets up a supervised learning problem to do unsupervised learning.
-Less connection.
CNN Supervised Image, sound, etc. -Need a large training dataset.
Autoencoder Unsupervised various -It can be trained in an unsupervised manner.
-It can be used for intrusion detection in the event of a poor reconstruction.
- Generating new content
- Filtering out noise