A mixture-of-expert VAE with Perceiver-style encoders learns a joint latent space for supernova light curves and spectra, enabling light-curve-to-spectrum generation that beats a contrastive nearest-neighbor baseline on simulated data.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
astro-ph.IM 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Mixture-of-Expert Variational Autoencoders for Cross-Modality Embedding of Type Ia Supernova Data
A mixture-of-expert VAE with Perceiver-style encoders learns a joint latent space for supernova light curves and spectra, enabling light-curve-to-spectrum generation that beats a contrastive nearest-neighbor baseline on simulated data.