Under a Locally Spike Sparse model and idealized Random Forest assumptions, LocalLSSFind provably recovers the signed features and interactions driving a single test prediction.
Iterative random forests to discover predictive and stable high-order interactions,
1 Pith paper cite this work, alongside 315 external citations. Polarity classification is still indexing.
1
Pith paper citing it
315
external citations · OpenAlex
fields
stat.ML 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Provable Recovery of Locally Important Signed Features and Interactions from Random Forest
Under a Locally Spike Sparse model and idealized Random Forest assumptions, LocalLSSFind provably recovers the signed features and interactions driving a single test prediction.