Efficient Intrusion Detection with KNN Classification and DS Theory - Proceedings of All India Seminar on Biomedical Engineering 2012 (AISOBE-2012) - page 178

Table 2 Result of tested

Category Data set 1 Data set 2

data set

Normal 680 673

Probs 44 46

DoS 118 121

R2R 100 103

R2L 57 57

Experimental Results and Performance Analysis

This section describes the experimental results and performance evaluation of the
proposed system. The proposed system is implemented in MATLAB (R2010a) and
the performance of the system is evaluated False positive rate (FP), False negative
rate (FN), True positive rate (TP), True negative rate (TN) ,and accuracy in
respect of true positive and true negative rate. For experimental evaluation, we
have taken KDD cup 99 data set, which is mostly used for evaluating the per-
formance of the intrusion detection system. Here, we have used only 1,000
instances of data of KDD Cup 99 data set for training and testing.
We have supervised different data sets with each 1,000 instances of data under
the result of ratio of attacks as represented in tabular format in Table 2 :

Performance Analysis

As seen from the output of performance on data sets it can be made out that when
KNN is combined with DS method, the performance gets significantly improved.
Earlier application of isolated KNN on data set has much greater accuracy than
later by integrating both KNN and DS Methods. Also, there is a considerable
enhancement in the true positive and true negative detection ratios and false
positive and false negative ratios. Thus, this gives the direct improvized accuracy in
the result. In this paper, we are showing the result for the parameters—Accuracy,
FP, FN, TP, TN only for one data set, i.e, for data set-1. Also, below is shown the
graph for that particular data set, and how to calculate these parameters with
suitable formulas.

correct detection
Total detection

FP =

FN ¼ Total detection false positive

TP þ TN
TP þ FP þ TN þ FN

Accuracy ¼