Hi-Fi decomposes the LOCO feature importance into unique, redundant, and synergistic components by finding feature subsets that minimize and maximize prediction-error reduction.
Barrett, Phys
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
physics.data-an 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Assessing high-order effects in feature importance via predictability decomposition
Hi-Fi decomposes the LOCO feature importance into unique, redundant, and synergistic components by finding feature subsets that minimize and maximize prediction-error reduction.