On the Importance of Architectures and Hyperparameters for Fairness in Face Recognition.
Rhea SukthankerSamuel DooleyJohn P. DickersonColin WhiteFrank HutterMicah GoldblumPublished in: CoRR (2022)
Keyphrases
- hyperparameters
- face recognition
- model selection
- cross validation
- bayesian inference
- closed form
- bayesian framework
- noise level
- support vector
- random sampling
- maximum likelihood
- prior information
- gaussian process
- em algorithm
- maximum a posteriori
- incremental learning
- gaussian processes
- sample size
- incomplete data
- face images
- missing values
- noisy images
- computer vision
- parameter settings
- grid search
- parameter space
- higher order
- principal component analysis
- noise reduction
- feature extraction
- feature selection