Classification of radio sources through self-supervised learning

Open Access

Table B.2

F1 scores for each class and the corresponding macro average.

Class	p ≤ 85%	Entire sub-set	p ≥ 99%
artifact	0.22	0.59	0.77
amorphous	0.31	0.67	0.92
bright core	0.00	0.63	0.85
head-tail	0.17	0.36	0.59
single lobe	0.22	0.48	0.83
centre-bright	0.40	0.73	0.91
c. p. ellipse	0.33	0.42	0.70
symmetric double	0.41	0.64	0.87
edge-bright	0.54	0.65	0.81
diffuse bent	0.40	0.47	0.59
structured bent	0.26	0.51	0.67
circular diffuse	0.53	0.66	0.80
Macro average	0.37	0.60	0.80

Notes. The values are compared across the entire subset, the subset with predicted class probabilities below 85%, and the subset with predicted class probabilities above 99%.

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.