- ConvNet Classification: Aimed at automated identification of lensing images in astronomical surveys.
- ConvNet Regression: For a non-linear regression on strong lensing images from the simulations.
- Auto-encoders (under development): Feature-based encoding and decoding networks used for filtering astronomical images.
- Generative Adversarial Networks (under development): Aimed at generating new images from convolutional networks trained from simulated images.
nesar/GravLensCNN
Implementation of 4 deep learning schemes for Gravitational lensing analysis.
Python