LSTM-MDNz improves photometric redshift point estimates and calibrated PDFs over a BNN baseline on HSC GalaxiesML, cutting outliers ~20% and enabling high-purity catalogs via z_conf.
J., Moustakas, J., Blanton, M
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
astro-ph.GA 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Enhancing Photometric Redshift Estimation for LSST with a Hybrid LSTM-Mixture Density Network
LSTM-MDNz improves photometric redshift point estimates and calibrated PDFs over a BNN baseline on HSC GalaxiesML, cutting outliers ~20% and enabling high-purity catalogs via z_conf.