Rethinking Attention-Model Explainability through Faithfulness Violation Test.
Yibing LiuHaoliang LiYangyang GuoChenqi KongJing LiShiqi WangPublished in: ICML (2022)
Keyphrases
- management system
- computational model
- probability distribution
- test data
- theoretical analysis
- linear model
- knowledge base
- em algorithm
- network model
- formal model
- simulation model
- experimental data
- statistical model
- theoretical framework
- parameter estimation
- real time
- input data
- mobile robot
- prior knowledge
- similarity measure
- high level
- genetic algorithm
- data sets