2/8

I spent this day watching one of the Stanford lectures detailing the workings of convolutional neural networks. I learned more about the different optimization algorithms. I've found one  which seems especially useful, algorithm which utilizes "momentum" - the program is slower to change based on smaller minutiae and will only veer towards a different conclusion when a much more significant amount of evidence is present. This seems to make sense for application towards the Hubble Telescope images as there is a lot of noise present in them because the resolution is heavily limited.