Novel FTLRNN with Gamma Memory for Short-Term and Long-Term Predictions of Chaotic Time Series

Table 6

Error norms variations for the real Sunspots time series and laser time series for FTLRNN model on testing data set.

Error norms	Monthly sunspots time series			Laser time series
	(𝐾=6 months ahead prediction)			(𝐾=10 step ahead prediction)
	Transfer function tanh			Transfer function tanh
	MSE	NMSE	𝑟	MSE	NMSE	𝑟
𝐿1 (tanh)	0.02016	0.36673	0.85915	0.006171	0.2174	0.9014
𝐿2 (tanh)	0.00554	0.1008	0.95528	0.009563	0.337	0.8302
𝐿3 (tanh)	0.00571	0.10387	0.94882	0.0279	0.98417	0.3038
𝐿4 (tanh)	0.00956	0.17388	0.91375	0.2430	1.18040	0.2430
𝐿5 (tanh)	0.01618	0.29439	0.85462	0.02168	0.7641	0.5541
𝐿∞ (tanh)	0.00653	0.11889	0.92482	0.01536	0.5393	0.7887