search

chengsoonong / crowdastro

Cross-identification of radio objects and host galaxies by applying machine learning on crowdsourced training labels.
MIT License
13 stars 1 forks source link

Interesting related astronomy papers #98

Closed jbanfield closed 8 years ago

jbanfield commented 8 years ago

Unsupervised learning Hocking, Geach, Davey & Sun (2015) http://arxiv.org/abs/1507.01589

Supervised learning Dieleman, Willett & Dambre (2015) http://adsabs.harvard.edu/cgi-bin/bib_query?arXiv:1503.07077

jbanfield commented 8 years ago

and references within

jbanfield commented 8 years ago

ATLAS-SWIRE cross-ids without RGZ data. It looks like they assume that all the radio components are in a line.

Fan et al. (2015) http://adsabs.harvard.edu/abs/2015MNRAS.451.1299F

MatthewJA commented 8 years ago

An interesting point is that previous ML applications in astronomy seem to use much closer objects than we are — our radio objects are much further away and less resolved. For this reason, I'm curious as to what the other RGZ student project is doing.

I wonder if Fan et al. have a catalogue? That would make a very interesting comparison.

jbanfield commented 8 years ago

I will email Fan to see what is available.

Added you to the wiki page so you can have access to the other projects.

MatthewJA commented 8 years ago

Thanks!

I might make an issue containing all the required reading, since there's a few different issues about that now.

MatthewJA commented 8 years ago

Looks like the other project might find some ways to extract useful radio features. With that in mind, I'm going to avoid doing much of that topic — I'm quite happy with the CNN as it stands, bar issue #99.

MatthewJA commented 8 years ago

Merged into #101.