search

mlr-org / mlr3pipelines

Dataflow Programming for Machine Learning in R
https://mlr3pipelines.mlr-org.com/
GNU Lesser General Public License v3.0
137 stars 25 forks source link

Using a PipeOp with a fixed state during training #537

Open mb706 opened 3 years ago

mb706 commented 3 years ago

This may be useful in some cases: use a PipeOp that is already trained inside a Graph, and use its predict() function during both training and prediction. Maybe this is useful when trying to get multiple tasks to the same scale / PCA rotation or something like this.

We could also solve this by having a PipeOpProxy-like construct that just calls param_set$values$graph$predict() on incoming data.

nipnipj commented 3 years ago

I think this also would help when using colapsefactors and editing its collapse_map.