Neural Net Noise Estimation

This project aims to construct and train a convolutional neural net which can estimate the level of Gaussian noise in any given Herschel observation.

To do this, I will create a number of fractal clouds using the ClusterFrac tool, and generate simple synthetic observations in each of the Herschel bands. To these, Gaussian noise will be added.

It is these observations that will be used to train the neural net.