Anomaly Detection Using Multiscale C-LSTM for Univariate Time-Series

Table 6

Comparisons of the four models in multiple dimensions includes evaluation metrics, running time and number of model parameters: the evaluation metrics include Recall (R), Precision (P), Accuracy (A) and F1.

Model	F1 (%)	R (%)	P (%)	A (%)	Time (s)	#parameters
C-LSTM-AE	96.2234	94.0161	98.7145	99.7347	8061	210076
IMC-LSTM	96.4273	94.8062	98.2382	99.7589	5113	150978
CMC-LSTM	96.8348	94.8677	98.9918	99.7759	3095	136930
SMC-LSTM	96.5864	94.5788	98.7831	99.7507	3448	46530

The bold values represent the best values of the evaluation metrics.