FRS targets backdoor triggers with MCTS-guided randomization and parameter smoothing, claiming a larger certified robustness radius that its own theory does not actually prove.
Fast and precise certification of transformers
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Certifying Language Model Robustness with Fuzzed Randomized Smoothing: An Efficient Defense Against Backdoor Attacks
FRS targets backdoor triggers with MCTS-guided randomization and parameter smoothing, claiming a larger certified robustness radius that its own theory does not actually prove.