A CNN autoencoder plus an off-manifold distance penalty is reported to improve detection of resolvable LISA sources in confusion-limited synthetic data (AUC 0.75 vs 0.56), but the weights were tuned on the test set.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
physics.gen-ph 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Manifold Learning for Source Separation in Confusion-Limited Gravitational-Wave Data
A CNN autoencoder plus an off-manifold distance penalty is reported to improve detection of resolvable LISA sources in confusion-limited synthetic data (AUC 0.75 vs 0.56), but the weights were tuned on the test set.