Rethinking Attention-Model Explainability through Faithfulness Violation Test.
Yibing LiuHaoliang LiYangyang GuoChenqi KongJing LiShiqi WangPublished in: CoRR (2022)
Keyphrases
- computational model
- high level
- parameter estimation
- database
- test data
- probabilistic model
- statistical model
- mathematical model
- theoretical framework
- theoretical analysis
- em algorithm
- probability distribution
- conceptual model
- multiscale
- autoregressive
- machine learning
- neural network
- management system
- sensitivity analysis
- pointwise
- data model
- prior knowledge
- objective function
- decision making
- learning algorithm
- data sets