Latin Hypercube Sampling does not yet accommodate categorical dimensions.

yngtdd / hyperspace

Distributed Bayesian Optimization

23 stars 8 forks source link

Open yngtodd opened 6 years ago

yngtodd commented 6 years ago

Latin hypercube sampling currently handlings integer and real valued dimensions.

Possible steps for categorical dimensions:

[ ] Create helper function that maps categorical dimensions to integer (based on index).
[ ] Handle categorical dimensions inside of lhs_start()