Probabilistic modelbased classification

Model-based classification is based on the specification of a probability distribution that describes the variability of the expression values. Typically, this is a mixture model, in which mixture components represent known classes (64). Model-based approaches are computation-intensive and can be sensitive to assumptions made about the probability model, but can provide a solid formal framework for the evaluation of many sources of uncertainty, and for assessing the probability of a sample belonging to a class.

0 0

Post a comment