A **supervised** algorithm is trained using data that is **labelled**, i.e. for which the question has already been answered that the model is designed to address. It then applies the insights gleaned during the training phase to new, unlabeled data in order to answer this question.

Supervised algorithms can be used in specific ways to reduce the effort required to label training data. **Semi-supervised** learning involves a training phase as supervised learning does, but not all the training data is labelled. Instead, it consists of a mixture of labelled and unlabelled data, and there is typically considerably more unlabelled than labelled data. **Active** learning is a special type of semi-supervised learning where the algorithm itself determines which training data the user is to label to achieve the best training results. This allows the algorithm to get the user to focus his labelling efforts on hard (=boundary) cases.

- used by
- ALG_Averaged one-dependence estimators ALG_Bayesian linear regression ALG_Bayesian network ALG_Convolutional neural network ALG_Decision tree ALG_Discriminant analysis ALG_Learning Vector Quantization ALG_Least Squares Regression ALG_Local regression ALG_Logistic regression ALG_Long short-term memory network ALG_Markov random field ALG_Multivariate adaptive regression splines ALG_Naive Bayesian Classifier ALG_Nearest Neighbour ALG_One Rule ALG_Perceptron ALG_Radial basis function network ALG_Random forest ALG_Stepwise Regression ALG_Support vector machine ALG_Zero Rule SPT_Bagging SPT_Boosting SPT_Elastic net SPT_Evolutionary selection SPT_LASSO SPT_Locally weighted learning SPT_Ridge regression SPT_Stacking