Understanding weight-magnitude hyperparameters in training binary networks.
Joris QuistYunqiang LiJan van GemertPublished in: ICLR (2023)
Keyphrases
- hyperparameters
- model selection
- cross validation
- random sampling
- bayesian inference
- bayesian framework
- support vector
- closed form
- em algorithm
- sample size
- grid search
- prior information
- gaussian process
- noise level
- maximum likelihood
- training set
- incremental learning
- gaussian processes
- maximum a posteriori
- incomplete data
- image processing
- regularization parameter
- noisy images
- random forest
- parameter settings
- bayesian networks