ML-Blink

V1.0 of the Citizen Science project is out!

MLblink_nov2019
(Screenshot ML-Blink tool)

Go to this website, load the data, toy with the tool, and find those intriguing strangers in existing sky surveys.

The overall challenge goes as follows: you’re presented with two images from the same location in the sky (i.e. a mission). Those two images are taken from two different sky surveys, USNO and PanSTARRS. We found those by working through the complete surveys using advanced computational tools. So each of the presented missions have an element of mystery. The overall idea is inspired from the blink comparator, used back in the 1900s to discover the oddity of Pluto.

Kristiaan Pelckmans och Beatriz Villarroel Rodriguez
(Copyright Johan Marklund)

However – as reality has it – those mystery findings are all too often a consequence of observational errors, or other complex irrelevancies. Existing autonomous astronomical pipelines have always (to some degree) a problem with those. At the end of the day, when scouting for interesting anomalies, human experts are indispensable.

That’s where you come in: we give you a pair of images which contains an element of mystery, and you try to figure out whether it’s really interesting or that this mission can be explained away using elementary matching techniques.

There is an added layer of complexity – there always is. We have an AI system in the back trying to learn your actions. Try to beat the AI system.  If the AI system gets really powerful, it will help you to go effectively through the observations: we serve you always the mission that the AI thinks has the largest potential. If you never bother to beat the AI, then the system will just walk randomly through the Sky.

Note that for various reasons, the AI-system works per-session. Your AI-buddy doesn’t talk to mine, or vice versa. We have no idea what your AI does or doesn’t. If you don’t like what it’s doing, you can just refresh your browser and reset what was learned.

Beatriz, Mikael, Abel, Diego, Johan, Kristiaan and many others (see team) have been working on this around the clock, out of pure enthusiasm. Join us! Learn a bit about deep-space astronomy and AI and you might find jus that bit excitement that escaped attention thus far.

So do the ML-Blink!

 

About | News&blogPhysics | Events | ML | Links | Research | Contact