Table 3.

Comparison of existing research based on dimensionality reduction in IDS.

Reference	Method Used	Major Contribution	Challenges/Limitations
[23]	Principal component analysis	Able to manage large datasets, more efficient.	Not able to handle nonlinear problems.
[24]	Auto-encoders	It does not require any prior assumptions for the reduction.	Slower in speed.
[25]	Missing value ratio	Mainly finds out the missing values and NULL values.	Works on the specific data framework.
[26]	Low variance filter	It eliminates the low variance filter in specific dimensions.	Can work on limited data.
[27]	Factor analysis	It can analyze various data factors.	Slower.
[28]	Forward feature selection	It works in the forward direction.	Works on the specific data framework.
[29]	Uniform manifold approximation and projection (UMAP)	UMAP is crafted from a theoretical foundation predicated on Riemannian manifolds and algebraic configuration.	Can work on limited data.
[30]	Random forest	It constructs a dimension reduction tree based on the decision.	Can work on limited data.