Treant trains accurate decision tree ensembles that are nearly insensitive to evasion attacks by minimizing an evasion-aware loss via robust splitting and attack invariance.
Combining email models for false positive reduction,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Treant: Training Evasion-Aware Decision Trees
Treant trains accurate decision tree ensembles that are nearly insensitive to evasion attacks by minimizing an evasion-aware loss via robust splitting and attack invariance.