Hands-On Artificial Intelligence for IoT

Hyperparameter tuning and grid search

Different models have different hyperparameters; for example, in linear regressor, the learning rate was a hyperparameter; if we're using regularization, then the regularizing parameter λ is a hyperparameter. What should be their value? While there's a rule of thumb for some hyperparameters, most of the time we make either a guess or use grid search to perform a sequential search for the best hyperparameters. In the following, we present the code to perform hyperparameter search in the case of SVM using the scikit library; in the next chapter, we'll see how we can use TensorFlow to perform hyperparameter tuning:

Grid search for best model and parametersfrom sklearn.model_selection import GridSearchCV ...