Spectral classification of young stars using conditional invertible neural networks - I. Introducing and validating the method

Open Access

Table 2

Average prediction performance of three networks (Settl-Net, NextGen-Net, and Dusty-Net) on 13 107 Phoenix synthetic models in the test set.

	RMSE			NRMSE

Network	log T_eff	log(ɡ)	A_V	log T_eff	log(ɡ)	A_V
Settl	4.260 × 10⁻⁴	1.211 × 10⁻²	7.893 × 10⁻³	9.904 × 10⁻⁴	4.846 × 10⁻³	7.893 × 10⁻⁴
NextGen	3.064 × 10⁻⁴	6.742 × 10⁻³	6.499 × 10⁻³	7.123 × 10⁻⁴	2.697 × 10⁻³	6.499 × 10⁻⁴
Dusty	7.274 × 10⁻⁵	1.573 × 10⁻³	2.517 × 10⁻³	3.888 × 10⁻⁴	7.863 × 10⁻⁴	2.517 × 10⁻⁴

Notes. For each parameter and each network, we present the RMSE, the mean accuracy of the MAP estimates, and the RMSE normalised by the parameter range covered in the training data (NRMSE). The test set of each network is drawn from the corresponding synthetic database.

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.