Unsupervised learning

Machine learning andData mining
Problems
Classification Clustering Regression Anomaly detection Association rules Reinforcement learning Structured prediction Feature learning Online learning Semi-supervised learning Grammar induction
Supervised learning (classification,regression)
Decision trees k-NN Linear regression Naive Bayes Logistic regression SVM Neural networks
Clustering
Hierarchical k-means EM DBSCAN OPTICS Mean-shift
Anomaly detection
k-NN Local outlier factor
Theory
Bias-variance dilemma Computational learning theory Empirical risk minimization PAC learning VC theory
Computer science portal Statistics portal
v t e

[hide]This article has multiple issues. Please help improve it or discuss these issues on thetalk page. This article may require cleanup to meet Wikipedia's quality standards. (May 2010) This article includes a list of references, butits sources remain unclear because it has insufficient inline citations. (February 2010)

In machine learning, the problem of unsupervised learning is that of trying to find hidden structure in unlabeled data. Since the examples given to the learner are unlabeled, there is no error or reward signal to evaluate a potential solution. This distinguishes unsupervised learning from supervised learningand reinforcement learning.

Unsupervised learning is closely related to the problem of density estimationin statistics.^[1] However unsupervised learning also encompasses many other techniques that seek to summarize and explain key features of the data. Many methods employed in unsupervised learning are based on data miningmethods used to preprocess^{[citation needed]} data.

Approaches to unsupervised learning include:

• clustering (e.g., k-means, mixture models, hierarchical clustering),
• hidden Markov models,
• blind signal separation using feature extraction techniques fordimensionality reduction (e.g., principal component analysis, independent component analysis, non-negative matrix factorization, singular value decomposition).^[2]

Among neural network models, the self-organizing map (SOM) and adaptive resonance theory (ART) are commonly used unsupervised learning algorithms. The SOM is a topographic organization in which nearby locations in the map represent inputs with similar properties. The ART model allows the number of clusters to vary with problem size and lets the user control the degree of similarity between members of the same clusters by means of a user-defined constant called the vigilance parameter. ART networks are also used for many pattern recognition tasks, such as automatic target recognition and seismic signal processing. The first version of ART was "ART1", developed by Carpenter and Grossberg (1988).^[3]

Bibliography[edit]

1. Jump up^ Jordan, Michael I.; Bishop, Christopher M. (2004). "Neural Networks". In Allen B. Tucker. Computer Science Handbook, Second Edition (Section VII: Intelligent Systems). Boca Raton, FL: Chapman & Hall/CRC Press LLC. ISBN 1-58488-360-X.
2. Jump up^ Acharyya, Ranjan (2008); A New Approach for Blind Source Separation of Convolutive Sources, ISBN 978-3-639-07797-1(this book focuses on unsupervised learning with Blind Source Separation)
3. Jump up^ Carpenter, G.A. and Grossberg, S. (1988). "The ART of adaptive pattern recognition by a self-organizing neural network". Computer 21: 77–88.

• Hinton, Geoffrey; Sejnowski, Terrence J. (editors) (1999); Unsupervised Learning: Foundations of Neural Computation,MIT Press, ISBN 0-262-58168-X (This book focuses on unsupervised learning in neural networks)
• Duda, Richard O.; Hart, Peter E.; and Stork, David G. (2001); Unsupervised Learning and Clustering, Chapter 10 inPattern classification (2nd edition), p. 571, New York, NY: Wiley, ISBN 0-471-05669-3
• Zoubin Ghahramani (September 16, 2004). "Unsupervised Learning".