Change training data and copy model parameters

GAMES-UChile / mogptk

Multi-Output Gaussian Process Toolkit

MIT License

161 stars 42 forks source link

Change training data and copy model parameters #32

Closed tdewolff closed 2 years ago

tdewolff commented 2 years ago

After training, be able to change the training data
Be able to save just the model parameters (not data) or copy to a new model

tdewolff commented 2 years ago

Is it possible to change the training data after training a model? Some models don't allow this (OpperArchambeau and Hensman, which keep q_mu and q_var parameters that have the same length as the input data), and I'm unsure if it makes sense.

Other than that, kernel parameters can now be copied using model.copy_parameters(other_model), see https://github.com/GAMES-UChile/mogptk/commit/eefbd8c5afb764d4491f1e41d22650ee2102e115

tdewolff commented 2 years ago

It's best to copy model parameters instead of changing the training data, closing