The triggers that open the NLP model backdoors are hidden in the adversarial samples.

Kun Shao Yu Zhang Junan Yang Xiaoshuai Li Hui Liu

Published in: Comput. Secur. (2022)

Keyphrases

high level
statistical model
hidden markov models
data sets
computational model
probabilistic model
formal model
network structure
test data
petri net
theoretical analysis
probability distribution
data model
objective function
knowledge base
feature selection
artificial intelligence