Automatic Gradient Descent: Deep Learning without Hyperparameters.
Jeremy BernsteinChris MingardKevin HuangNavid AzizanYisong YuePublished in: CoRR (2023)
Keyphrases
- deep learning
- hyperparameters
- model selection
- cross validation
- closed form
- bayesian framework
- unsupervised learning
- machine learning
- maximum likelihood
- support vector
- prior information
- random sampling
- gaussian process
- bayesian inference
- sample size
- em algorithm
- noise level
- weakly supervised
- image processing
- maximum a posteriori
- mental models
- parameter settings
- incremental learning
- parameter space