OutFlip: Generating Out-of-Domain Samples for Unknown Intent Detection with Natural Language Attack.
DongHyun ChoiMyeongcheol ShinEungGyun KimDong Ryeol ShinPublished in: CoRR (2021)
Keyphrases
- natural language
- detection algorithm
- domain specific
- attack detection
- domain independent
- data sets
- detection method
- object detection
- knowledge representation
- domain dependent
- automatic detection
- machine learning
- detection mechanism
- semantic interpretation
- false alarms
- detection rate
- false positives
- cross domain
- detection accuracy
- language processing
- generation process
- medical domain
- active learning
- question answering
- anomaly detection