MockingBERT: A Method for Retroactively Adding Resilience to NLP Models.
Jan JezabekAkash SinghPublished in: CoRR (2022)
Keyphrases
- experimental evaluation
- probabilistic model
- natural language processing
- statistical model
- high accuracy
- high precision
- computational complexity
- parameter estimation
- mathematical model
- prior knowledge
- computational cost
- pairwise
- preprocessing
- bayesian networks
- statistical methods
- detection method
- em algorithm
- machine learning methods
- information extraction
- support vector machine svm
- model selection
- input data
- neural network
- classification accuracy
- objective function
- similarity measure
- genetic algorithm