Breaking BERT: Understanding its Vulnerabilities for Named Entity Recognition through Adversarial Attack.
Anne DirksonSuzan VerberneWessel KraaijPublished in: CoRR (2021)
Keyphrases
- named entity recognition
- information extraction
- natural language processing
- named entities
- text summarization
- attack graphs
- conditional random fields
- semi supervised
- countermeasures
- maximum entropy
- attack graph
- hand coded
- sequence labeling
- annotated corpus
- relation extraction
- classifier ensemble
- security vulnerabilities
- text mining
- natural language
- artificial intelligence
- knowledge discovery
- proper names
- buffer overflow
- information security