Image Details
Caption: Figure 2.
Network architectures for the VMP identification framework. Panel (a) illustrates the 1D-ResNet classifier used for identifying VMP candidates, which outputs a probability distribution. Panel (b) shows the regression network designed to estimate three continuous atmospheric parameters (Teff, ﹩\mathrm{log}g﹩, and [Fe/H]). Both models utilize residual blocks to extract features effectively from the one-dimensional spectra.
© 2026. The Author(s). Published by the American Astronomical Society.