1. Machine learning uses algorithms to learn patterns from labeled data samples represented as feature vectors. Models like neural networks, support vector machines, decision trees, and boosting algorithms are used to classify patterns. 2. Neural networks use layers of nodes and backward propagation of errors to learn weights, while support vector machines find an optimal separating boundary between classes. Decision trees use feature testing and impurity measures to split data into branches. 3. Bayesian approaches apply probability densities and maximum entropy principles to compute posterior distributions over classes given observed data and features.