To overcome these limitations, the researchers implemented a deep learning approach using a Convolutional Variational Autoencoder (VAE) for compressing Ca II (8542 Å) spectral data.