Bypassing the static input size of neural networks in flare forecasting by using spatial pyramid pooling

Home

All issues

Volume 695 (March 2025)

A&A, 695 (2025) A65

Full HTML

Open Access

Table B.2.

Same as B.1 but using the MX classification.

Classification	Model	Loss	Class	Label	Correctly Predicted	Wrongly Predicted	Acc^a	Acc Std^b
MX	Trad-CNN	BCE	1	X	19	7	0.73	0.38
				M	158	86	0.66	0.13

			0	C	806	791	0.5	0.06
				NF	7811	885	0.9	0.09
				FX	12	11	0.52	0.48
				FM	219	70	0.75	0.16
				FC	2035	564	0.78	0.06

		TSS	1	X	15	11	0.71	0.35
				M	227	60	0.79	0.12

			0	C	564	920	0.37	0.22
				NF	5563	3445	0.63	0.27
				FX	28	41	0.44	0.19
				FM	135	89	0.51	0.28
				FC	1330	1281	0.51	0.24

	SPP-CNN	BCE	1	X	21	5	0.78	0.39
				M	195	75	0.69	0.39

			0	C	853	695	0.53	0.3
				NF	6780	1948	0.79	0.23
				FX	28	5	0.24	0.42
				FM	189	80	0.71	0.27
				FC	1608	883	0.65	0.3

		TSS	1	X	19	7	0.6	0.49
				M	154	99	0.61	0.32

			0	C	950	542	0.64	0.21
				NF	7671	772	0.91	0.07
				FX	15	6	0.41	0.43
				FM	271	137	0.68	0.28
				FC	1918	535	0.79	0.16

^(a)

Accuracy

^(b)

Accuracy Standard deviation

Notes. Value underlined highlight the best result among all models and Loss function.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.