Pattern recognition

Pattern recognition is the process of identifying regularities and meaningful structures in information. It can be performed by people, animals, or computational systems.

Feature extraction: The transformation of raw data into measurable attributes useful for analysis. Recognition systems rely on informative attributes to distinguish patterns.
Feature space: A mathematical space whose dimensions represent measurable attributes of data. A pattern’s position in this space determines which observations appear similar.
Optical character recognition: The conversion of images of printed or handwritten text into machine-readable characters. It identifies letter patterns in images and turns them into text.
Statistical pattern recognition: The use of statistical models and decision rules to assign observations to patterns or classes. It is a formal approach to recognition that represents uncertainty mathematically.
Classification: The assignment of observations to categories according to their features or a learned model. It is a common recognition task when possible patterns belong to known classes.
Similarity measure: A function that quantifies how alike two objects or data points are. Recognition often depends on deciding which stored patterns are closest.
Speech recognition: The conversion of spoken language into text or machine-interpretable representations. It detects linguistic patterns in acoustic signals.
Structural pattern recognition: The recognition of patterns through relationships among their constituent parts. It represents structure explicitly rather than relying only on numerical feature similarity.
Clustering: The grouping of data points by similarity without requiring predefined class labels. It reveals recurring structures when categories are not known in advance.
Training data: Examples used to fit a machine-learning model or guide a learning process. Supervised recognizers learn category boundaries from labeled examples.

Linked from 12 pages