A Taxonomy of Machine Learning Algorithms and Learning Modes
Summary
This overview organizes common machine learning methods in two ways: by their form or function, and by how they learn from data. It surveys regression, instance-based methods, regularization, decision trees, Bayesian methods, clustering, association rules, neural networks, deep learning, dimensionality reduction, and ensembles. It gives representative algorithms for each group and briefly describes the purpose of methods such as clustering and dimensionality reduction.
The second classification distinguishes supervised learning with labeled examples, unsupervised learning that seeks structure in unlabeled data, and semi-supervised learning that combines labeled and unlabeled examples. The article connects these ideas to quantitative trading, where models may analyze market data or produce signals, but it does not provide a trading strategy, implementation guidance, or empirical results. The lists are explicitly incomplete, categories can overlap, and some terms describe either a method or a problem, so the taxonomy is best treated as an introductory map rather than a definitive classification.
Key ideas
- Machine learning methods can be grouped by function or by the way they learn from data.
- Regression, trees, Bayesian methods, clustering, neural networks, dimensionality reduction, and ensembles address different modeling tasks.
- Supervised learning trains on labeled examples, while unsupervised learning looks for structure without target labels.
- Semi-supervised learning uses both labeled and unlabeled data.
- Algorithm categories overlap, and the article presents a selective overview rather than a complete taxonomy.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.