Skip to content
No description, website, or topics provided.
Python
Branch: master
Clone or download
Cannot retrieve the latest commit at this time.
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
input_imgs
BGMM.py
CAE.py
README.md
pretrained_CAE_model.h5
pretrained_CAE_reconstruction.png

README.md

Unsupervised-ML-on-Strong-Gravitational-Lensing-Detection-using-Convolutional-Autoencoder

This toy code is simplified for demonstration based on the codes used in the paper: Identifying Strong Lenses with Unsupervised Machine Learning using Convolutional Autoencoder, Cheng et al. 2019".

The goal of this is to apply an unsupervised machine leanring techniques consisting of Convolutional Autoencoder and Bayesian Gaussian Mixture Model to identify galaxy-galaxy strong lensing systems.

  • Feature Extraction using Convolutional Autoencoder (CAE.py)
  • Clustering data at the high-dimensional feature space by Bayesian Gaussian Mixture model (BGMM.py)

100 images attached in this repo are in linear scale and after the denoise process by the CAE with a simplified architecture (no dense layer, details are shown in paper attached above). The complete original simulated data used and the classification table can be found in Strong Lenses Finding Challenge v1.

You can’t perform that action at this time.