Characterising Gradients for Unsupervised Accuracy Estimation under Distribution Shift.
Renchunzi XieAmbroise OdonnatVasilii FeofanovIevgen RedkoJianfeng ZhangBo AnPublished in: CoRR (2024)
Keyphrases
- high accuracy
- estimation accuracy
- probability distribution
- computational efficiency
- probability density
- estimation error
- supervised learning
- prediction accuracy
- computational cost
- parameter estimation
- unsupervised learning
- classification accuracy
- error rate
- random variables
- data driven
- density estimation
- highly accurate
- least squares
- error analysis
- weakly supervised
- computational complexity
- unsupervised manner
- maximum likelihood estimator