R-LOCO improves local feature attributions by dividing the feature space into regions with consistent importance patterns and applying global attribution methods regionally.
By definition, the global piecewise model f(x) = Pm k=1 fk(x)1x∈Ak simplifies to the active local model for any point within its region
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Regional Explanations: Bridging Local and Global Variable Importance
R-LOCO improves local feature attributions by dividing the feature space into regions with consistent importance patterns and applying global attribution methods regionally.