Research Article

Trust in Intrusion Detection Systems: An Investigation of Performance Analysis for Machine Learning and Deep Learning Models

Table 17

The performance of deep learning for KDD dataset using LSTM and GRU with selected features.

ModelEvaluation metricCross-validation performanceTesting performance
DOSNormalProbeR2LU2RDOSNormalProbeR2LU2R

LSTM with one layerTNR66.9698.7510099.9910078.0698.50100.0099.99100.00
FPR33.041.2500.01021.941.500.000.010.00
FNR0.9530.6310079.181001.0619.57100.0077.13100.00
Accuracy92.9893.5199.1399.7999.9994.9895.2799.1599.7999.99
Precision92.9192.35084.040.095.0692.090.082.110.0
Recall99.0569.37020.82098.9480.430.0022.870.00
F-score95.8477.590.039.480.096.9685.870.035.780.0

LSTM with two layerTNR81.5398.3799.9910010085.8398.37100.0099.99100.00
FPR18.471.630.010014.171.630.000.010.00
FNR1.2415.9387.6994.941001.4011.5076.0976.83100.00
Accuracy95.4995.8299.2299.7699.9996.1796.6099.3599.7999.99
Precision95.8291.7884.5891.370.096.7492.1998.1886.810.0
Recall98.7684.0712.315.06098.6088.5023.9123.170.00
F-score97.2687.7133.9439.520.097.6690.3138.4636.570.0

GRU with one layerTNR78.7399.2210010010079.0099.19100.00100.00100.00
FPR21.270.7800021.000.810.000.000.00
FN0.3618.261001001000.3618.15100.00100.00100.00
Accuracy95.6896.1199.1399.7599.9995.7296.0999.1599.7499.99
Precision95.2595.770.00.00.095.2995.660.00.00.0
Recall99.6481.7400099.6481.850.000.000.00
F-score97.488.20.00.00.097.4288.220.00.00.0

GRU with two layerTNR82.998.9499.9999.9810083.2599.3499.9899.99100.00
FPR17.11.060.010.02016.750.660.020.010.00
FNR0.7714.486.8373.341000.4614.1176.0976.54100.00
Accuracy96.1496.5699.2499.899.9996.4596.9399.3499.7999.99
Precision96.1394.6291.6780.40.096.2096.5993.0883.330.0
Recall99.2385.613.1726.66099.5485.8923.9123.460.00
F-score97.6589.8835.2839.920.097.8490.9238.0536.610.0