Where to Attack: A Dynamic Locator Model for Backdoor Attack in Text Classifications.
Heng-Yang LuChenyou FanJun YangCong HuWei FangXiao-Jun WuPublished in: COLING (2022)
Keyphrases
- computational model
- prior knowledge
- theoretical analysis
- experimental data
- information retrieval
- high level
- probabilistic model
- em algorithm
- dynamic environments
- theoretical framework
- dynamically updated
- neural network
- detection mechanism
- countermeasures
- formal model
- language model
- genetic algorithm
- machine learning