In contrast to learned parameters, hyper-parameters cannot be directly learned from the training process. In the context of ML, hyper-parameter setting can be seen as model selection. Examples of hyper-parameters in ML models include: