Skip to main content
Search

# ExplainReduce

1 related articles

Subscribe to updates
  1. ExplainReduce cuts hundreds of local explanations to five, turning proxy models into a readable global explanation

    Post-hoc local explanations are numerous and redundant, making them hard to read as a whole. ExplainReduce casts the choice of representative local explanations as a set-cover-like submodular maximization solved by greedy algorithms with worst-case approximation guarantees. Across twelve synthetic and real datasets, a proxy set of as few as five local explanations approaches the fidelity of a full set of 500 on both training and test data, and works for regression and classification with several local explanation methods.

You’re all caught up