[Retracted] Transfer Learning Auto-Encoder Neural Networks for Anomaly Detection of DDoS Generating IoT Devices

Table 3

Percentage decrease in model accuracy before and after transfer-learning (TL) across device and malware type.

IoT device	Mirai to Bashlite (% decrease)		Bashlite to Mirai (% decrease
	Pre-TL	Post-TL	Pre-TL	Post-TL
DAD	34.66	10.09	49.14	−0.33
ECT	28.68	1.45	32.76	−0.62
P737	33.30	1.78	37.19	−2.11
P838	30.26	4.87	34.42	−0.59
S1003	33.36	5.12	27.14	−1.95
S1002	37.97	1.37	24.00	0.36
PBM	29.03	2.46	13.28	−0.85
SNH	—	—	22.16	−1.56
END	—	—	48.04	10.36
Average (%)	32.47	3.88	32.02	0.30