search

tallamjr / astronet

Efficient Deep Learning for Real-time Classification of Astronomical Transients and Multivariate Time-series
Apache License 2.0
14 stars 3 forks source link

[feature/20/plasticc] Investigate options for Gaussian Process regression for PLAsTiCC data #21

Closed tallamjr closed 4 years ago

tallamjr commented 4 years ago

Linked to feature #20

SPCC PLAsTiCC data is irregularly sampled, but one required a regular sampling of points for the t2 algorithm to work best. The goal would be to use gaussian process regression to interpolate between the points in 6 dimensions.

GPFlow seems to be a good candidate for this

Update

It was determined that initially working with PLAsTiCC data was easier for experimentation and prototyping, so this issue has moved to working with that dataset instead for now.

tallamjr commented 4 years ago

Now using george and code snippets from Catarina

tallamjr commented 4 years ago

Closed with #23