Table 7.

Performance metrics (in percent) of different feature extraction models on the OPPORTUNITY dataset with different window sizes T.

2*Method	T = 32			T = 64 (Baseline)			T = 96
	Acc.	WF1	AF1	Acc.	WF1	AF1	Acc.	WF1	AF1
HC	88.74	88.32	58.93	89.96	89.53	63.76	90.38	89.98	64.87
CBH	88.99	88.23	58.20	89.66	88.99	62.27	89.30	88.81	61.99
CBS	89.61	89.07	63.72	90.22	89.88	67.5	90.36	89.96	66.92
MLP	90.79	90.4	66.33	91.11	90.86	68.17	90.94	90.65	66.37
CNN	90.34	89.71	62.10	90.58	90.19	65.26	90.38	89.98	63.38
LSTM	90.88	90.60	67.20	91.29	91.16	69.71	91.33	91.21	68.64
Hybrid	91.10	90.75	67.31	91.76	91.56	70.86	91.44	91.25	69.04
AE	85.88	85.90	49.54	87.80	87.60	55.62	86.84	86.78	52.67