New maxNrAssignment
scheme for pruning
#1156
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
As discussed in our meeting yesterday, this PR updates the
DecisionTreeFactor::prune
method to use the newmaxNrAssignments
scheme.Here, we use the record of assignments for each leaf to compute the number of duplicate values in the tree without enumerating all the (exponential) number of assignments, and threshold on the top
maxNrAssignments
values. To facilitate this, I added a newvisitLeaf
method toDecisionTree
which follows the same idea asvisit
in that we visit all the leaves (and not assignments), butvisitLeaf
passes the full leaf object to the functional, allowing us to doleaf.nrAssignments()
andleaf.constant()
quite easily.NOTE I have one question: should the functional receive the leaf object or the leaf pointer? It is a const either way, but I've made it as a leaf object const-ref so there shouldn't be a memory hit and it is convenient to use (since we don't declare a
LeafPtr
typedef).