Table 1.
Type of ML | Mechanism | Type of Data Provided | Problems that Can Solve | Examples of Models |
---|---|---|---|---|
Supervised learning (SL) | The algorithm is provided with tuples of input and output (x,y) and the algorithms infer the relation that maps the dataset | Labeled data | Classification task (discrete variable) Regression task (continuous variable) |
Logistic regression Decision Tree Random Forest ANNs |
Unsupervised learning (UL) | The algorithm exhibits self-organization to capture hidden patterns in data | Unlabeled data | Clustering Association Anomalies detection |
Hierarchical clustering K-mean DBSCAN ANNs |
Semi-supervised learning (SSL) | Falls between unsupervised learning (with no labeled training data) and supervised learning (with only labeled training data; a mix of SL and UL | Mostly unlabeled data, with a small amount of labeled data | Transductive task (infer the correct labels for the given unlabeled data) or inductive tasks (infer the correct mapping from x to y). | Generative model Self-training model Co-training model Transductive model Graph-based model |
Reinforcement learning (RL) | The algorithm is programmed with a goal and a set of rules. It tends to a nearly optimal policy that maximizes the “reward function” or reinforcement signals | Not needing labeled input/output pairs to be presented, only a numerical performance score is given as guidance. | Economics and game theory under bounded rationality, control theory | Monte Carlo methods Q-learning SARSA methods |