Goal-Oriented Sensitivity Analysis of Hyperparameters in Deep Learning.
Paul NovelloGaël PoëtteDavid LugatoPietro Marco CongedoPublished in: J. Sci. Comput. (2023)
Keyphrases
- goal oriented
- sensitivity analysis
- deep learning
- hyperparameters
- model selection
- cross validation
- bayesian inference
- unsupervised learning
- closed form
- random sampling
- bayesian framework
- support vector
- prior information
- em algorithm
- sample size
- machine learning
- gaussian process
- maximum likelihood
- incremental learning
- noise level
- maximum a posteriori
- managerial insights
- incomplete data
- parameter settings
- variational inequalities
- parameter space
- weakly supervised
- missing values
- mental models
- training set
- pattern recognition
- data sets
- active learning
- prior knowledge
- image segmentation
- image processing
- computer vision