Recently, Anna Scaife released HTRU1 dataset which was originally used to train SPINN pulsar classifier. She also shared code for classification of pulsar and non-pulsar using CNN in Pytorch. I guess can integrate the analysis with MiraPy since the structure of dataset is very similar to MNIST dataset so we can look into other applications of Deep Learning.
The first and foremost task would be writing a DataGenerator class using keras. The neural network implementation has a scope of improvement through data augmentation. We would look into other possible ways to create the 'state of the art' technique for the same. This is a really good problem for first timers.
Recently, Anna Scaife released HTRU1 dataset which was originally used to train SPINN pulsar classifier. She also shared code for classification of pulsar and non-pulsar using CNN in Pytorch. I guess can integrate the analysis with MiraPy since the structure of dataset is very similar to MNIST dataset so we can look into other applications of Deep Learning.
The first and foremost task would be writing a
DataGenerator
class usingkeras
. The neural network implementation has a scope of improvement through data augmentation. We would look into other possible ways to create the 'state of the art' technique for the same. This is a really good problem for first timers.Anna's tweet Her Notebook