Machine Learning

Learning problems

Supervised Learning: involves describing a class of problem through a model for learning a mapping between inputs and targets. There are two main types of supervised learning – a classification that involves predicting a class label and regression that involves predicting a numerical label.

Unsupervised Learning: this also involves describing a class of problems through a model but to describe or relate data. This also has its two main types – clustering that involves finding groups in data and density estimation that involves summarizing the distribution of data.

Two additional unsupervised methods are visualization and projection.

Reinforcement Learning: involves describing a class of problem where there is an agent operating in an environment and needs to learn with the use of feedback.

Hybrid Learning Problems

Semi-Supervised Learning: it is basically a mode of supervising d learning where the data used for training contains a lot of unlabeled examples and a few labeled ones.
Self-Supervised Learning: it basically involves an unsupervised learning problem which is framed as a supervised learning problem so that supervised learning algorithms can be applied to it and solved.
Multi-Instance Learning: this basically involves a supervised learning problem with unlabeled examples and labeled bags or groups of samples.

Statistical learning

Inductive Learning: it is basically the involvement of evidence to determine outcomes.
Deductive Inference: it is the involvement of general rules to determine outcomes.
Transductive Learning: this involves referring to specific examples from specific domains to make predictions.

Learning Techniques

Multi-Task Learning: basically, fits a model on one dataset addressing multiple related problems.
Active Learning: involves the model to query a human user while learning to resolve ambiguity during learning.
Online Learning: this involves using data available and updation of the model before a required prediction or after a prediction is made.
Transfer Learning: involves training of a model primarily on one task and then usage of some or all of the model as the initial point for a related task.
Ensemble Learning: involves fitting of one or more modes on the same data and combination of predictions from each model.

Discussion about the types of machine learning