Tuning hyperparameters of a SVM-based water demand forecasting system through parallel global optimization.
Antonio CandelieriIlaria GiordaniFrancesco ArchettiKonstantin BarkalovIosif B. MeyerovAlexey PolovinkinAlexander SysoyevNikolai Yu. ZolotykhPublished in: Comput. Oper. Res. (2019)
Keyphrases
- global optimization
- hyperparameters
- demand forecasting
- support vector
- water resources
- model selection
- cross validation
- historical data
- closed form
- bayesian inference
- bayesian framework
- random sampling
- parameter settings
- em algorithm
- particle swarm optimization
- regularization parameter
- maximum a posteriori
- sample size
- noise level
- incomplete data
- supply chain
- prior information
- pso algorithm
- maximum likelihood
- water quality
- forecasting accuracy
- incremental learning
- support vector machine
- svm classifier
- ls svm
- active learning
- support vector machine svm
- kernel function
- regression model
- missing values
- multi objective
- prior knowledge
- artificial neural networks