Algorithm 1 (already implemented in partial_fit) needs to be extended to take into account instance-based pruning, which aims to reduce the number of classifiers to a subset of size k << K needed for prediction while ensuring convergence to the same result as if using K classifiers.
This should be implemented in `CostSensitiveWeightedEnsemble::partial_fit`` (at the end of this function). The code should follow what has been described in Algorithm 2 in the paper.
Algorithm 1 (already implemented in
partial_fit
) needs to be extended to take into account instance-based pruning, which aims to reduce the number of classifiers to a subset of size k << K needed for prediction while ensuring convergence to the same result as if using K classifiers.