LiMA reformulates attribution as submodular subset selection and uses bidirectional greedy search to identify minimal important regions, reporting 36.3% better insertion and 39.6% better deletion scores than prior methods on eight foundation models.
Interpreting object-level foundation models via visual precision search,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Less is More: Efficient Black-box Attribution via Minimal Interpretable Subset Selection
LiMA reformulates attribution as submodular subset selection and uses bidirectional greedy search to identify minimal important regions, reporting 36.3% better insertion and 39.6% better deletion scores than prior methods on eight foundation models.