Spectral Deblending via Machine Learning
My primary research focuses on separating the spectral components of strongly lensed quasars from their host galaxies and the intervening lens. Using PyTorch and GRU (Gated Recurrent Units) networks, I develop algorithms for spectral decomposition that outperform traditional template-fitting methods.
- Impact: Improving time-delay measurements for Hubble constant determination in large-scale surveys.