The topic of this blog post is my project at Insight Data Science, a program that helps academics, like myself, transition from academia into industry. So as you have probably figured, I am looking for a job, so feel free to get in touch if you think I might be of interest to your company (in the US). The first part of the blog will be a high-level description of the data science. The specifics of the project, including code and low-level technical aspects, are treated in a second part.
Mixture Density Networks with Edward, Keras and TensorFlow
In the previous blog post we looked at what a Mixture Density Network is with an implementation in TensorFlow. We then used this to learn the distance to galaxies on a simulated data set. In this blog post we'll show an easier way to code up an MDN by combining the power of three python libraries.
Mixture Density Networks for Galaxy distance determination in TensorFlow
In this blog post I will explain a problem we encounter in observational cosmology called photometric redshifts and how we can use Mixture Density Networks (MDN's) to solve them with an implementation in TensorFlow. MDN's are just a different flavour of Neural Network. MDN's in the paper (PDF) by Bishop are applied to a toy problem trying to infer the position of a robotic arm. In this blog post I wanted to show the usage of MDN's on a real world problem, and with a real world problem I mean a simulated galaxy data set. The code used in this work is based on the second