Cybersecurity

Table 7 Experimental results of autoencoder with and without optimization

From: Quantized autoencoder (QAE) intrusion detection system for anomaly detection in resource-constrained IoT devices using RT-IoT2022 dataset

Methods	Error metric	Threshold	Accuracy (%)	Precision (%)	Recall (%)	F1 score (%)
Autoencoder	MSE	0.006	91.04	91.04	91.05	95.1
Autoencoder	MAE	0.02	98.40	98.39	98.40	98.39
QAE-f16	MSE	0.073	78.07	78.07	78.69	86.10
QAE-f16	MAE	0.145	97.25	97.24	97.25	97.24
QAE-u8	MSE	6153.01	92.08	92.10	92.08	95.10
QAE-u8	MAE	35.02	96.35	96.35	96.36	98.10

Back to article page