Multi-Criteria Feature Selection Based Intrusion Detection for Internet of Things Big Data

. 2023 Aug 25;23(17):7434. doi: 10.3390/s23177434

Algorithm 2: M-LVW feature selection algorithm

Input: Performance evaluation metric e (evaluation criterion for feature subsets) of the classifier; dataset D; feature set F; classification algorithm h; and stop condition control parameter T
Output: feature subset

F^{*}

Process:
1.

Initialize : s c o r e = CrossValidation (h (F, D), e)

; d = |F|

; F^{*} = F

; t = 0

2. If the value of evaluation metric e for the classifier is positively correlated with the performance of the classifier, then
3.

p = True

4. else
5.

p = False

6. end if
7. while

t < T

do
8.

Randomly generate feature subset F^{'}

d^{'} = |F^{'}|

10.

s c o r e^{'} = CrossValidation (h (F^{'}, D), e)

11.

\begin{array}{l} i f ((p = = True) AND (s c o r e^{'} > s c o r e)) OR \\ ((p = = False) AND (s c o r e^{'} < s c o r e)) OR \\ (s c o r e^{'} = = s c o r e AND d^{'} < d) then \end{array}

12.

F^{*} = F^{'}

; t = 0

; s c o r e = s c o r e^{'}

; d = d^{'}

13. else
14.

t = t + 1

15. end if
16. end while
return

feature subset F^{*}