Inferring properties of dust in supernovae with neural networks

Open Access

Table 3

Comparison of neural network performance for classifying dust species, and the fraction of predictions of the test data set that have to all the predictions from the test data set , in different scenarios for four different cases.

		Classification accuracy rate (%)
Case	Scenario	Carbon dust	Mixed dust	Silicate dust	(%)
	S1 ^(a)	86		95	100	−
1	S2 ^(b)	74		85	99	−
	S3 ^(c)	72		75	98	−
	S1 ^(a)	97		99	100	68
2	S2 ^(b)	98		99	100	31
	S3 ^(c)	97		94	100	12
	S1 ^(d)	89		90	98	−
3	S2 ^(e)	73		79	98	−
	S3 ^(f)	61		63	96	−
	S1 ^(d)	98		100	100	48
4	S2 ^(e)	94		95	99	34
	S3 ^(f)	95		57	99	07

Notes. The definition of cases are the same as in Table 2. With the subset of JWST filters that are selected via the feature selection procedure as follows: ^{(a) − (f)}: see the definitions in Table 2.

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.