We have a set of MiraPy tutorials for problem-solving in Astronomy using Deep Learning. You can find the Jupyter notebooks in our Github repository. Following are the short descriptions of MiraPy applications:
Astronomical Image Reconstruction using Autoencoder
Encoder-decoder networks can be trained for noise removal from blurry image. We can use MiraPy for astronomical image reconstruction by training a simple denoising autoencoder using some images of galaxies and nebulae in Missier catalog.
ATLAS variable star Classification
We demonstrate how to use MiraPy to classify variable stars using features extracted from light curves. These features are available in ATLAS catalog. We use deep neural network for the same.
OGLE variable star Classification
We demonstrate how to use MiraPy to classify variable stars using light-curves available in OGLE variable star catalogs. We use Recurrent Neural Network (RNN) in the classification model.
HTRU1 Batched Dataset Classification
MiraPy can be used for the classification of pulsars and non-pulsars in dataset released by HTRU1 survey. The dataset contains 60000 images which are classified using Convolutional Neural Network (CNN).
X-Ray Binary Classification
Tutorial demonstrates how to use Fully-Connected Neural (FCN) network to classify features of pulsar, non-pulsar and black hole systems.
2D and 3D visualisation
We demonstrate how to use MiraPy to visualize a feature dataset using 2D and 3D graphs. For this purpose, we use Pricipal Component Analysis (PCA) for feature reduction.