Implemented the following changes to active_learning.py that we discussed a couple of weeks ago:
Simplified the possession of acquisition_function and gl.ssl objects in the active_learner object. This makes things easier to use and less confusing what controls what--everything is controlled by the active_learner object.
Made proportional sampling easier to use in the active learning process, with capability to return the acquisition function values when selecting query points to allow for (i) easier visualizations in toy examples and (ii) other evaluations in experiments.
Implemented the following changes to
active_learning.pythat we discussed a couple of weeks ago:acquisition_functionandgl.sslobjects in theactive_learnerobject. This makes things easier to use and less confusing what controls what--everything is controlled by theactive_learnerobject.