Google Research releases exoplanet discovery ML code

This week, Google Research released the AI code used in the discovery of two new exoplanets. Their article includes an accessible intro to planet hunting and machine learning for those interested in either topic.

The associated paper is mostly about the AI but includes a good sense of just what real planetary research looks like and how far it’s come: Shallue, C.J. & Vanderburg, A., 2017. Identifying Exoplanets with Deep Learning: A Five Planet Resonant Chain around Kepler-80 and an Eighth Planet around Kepler-90. The Astronomical Journal, 155(2), p.94.

There’s an interview with Chris Shallue, the lead author on the paper on the TWIMLAI podcast which makes interesting listening for anyone interested in machine learning, AI or exoplanetary research.

Slaughtering Pixies

Well, it’s been quite an interesting couple of weeks. One of the things I am doing this week is to close down the Scottish Physics Teaching Resources site (, which for the past few years has provided a vehicle for Physics teachers in Scotland to share all kinds of teaching resources. It has been useful in a time of badly conceived and implemented curriculum change in Scotland.

Why close it down? Two reasons. Until very recently, I was intending to leave the country and work overseas and I was uncertain that I could sustain the required commitment to keep the site functioning well. Plans have changed (and continue to do so on a daily basis) but the community of teachers needs something reliable to help them deliver a workable curriculum. The second reason is that I need to remove as many distractions as possible to enable me to pick up a project I have been trying to progress for the past two years, without having the time to do so satisfactorily. Like the overseas project, it may yet come to nothing but a learning experience, but I have to give it my best shot.