Finite Neural Networks as Mixtures of Gaussian Processes: From Provable Error Bounds to Prior Selection.
Steven AdamsAndrea PatanèMorteza LahijanianLuca LaurentiPublished in: CoRR (2024)
Keyphrases
- gaussian processes
- error bounds
- dirichlet process
- bayesian methods
- gaussian process
- theoretical analysis
- gaussian process regression
- worst case
- multi task learning
- prior information
- human pose estimation
- preference learning
- prior knowledge
- covariance function
- hyperparameters
- mixture model
- multi task
- bayesian framework
- machine learning
- gaussian process models
- model selection
- probabilistic model