TOPICS
Search

Sequence Labeling


Sequence labeling is the assignment of one label y_t to each member x_t of an observed sequence x=(x_1,...,x_T). In a probabilistic formulation, a labeling method chooses the label sequence y=(y_1,...,y_T) that maximizes a conditional probability,

 y^^=argmax_(y)P(y|x).

Unlike independent classification of the individual observations, sequence labeling can model dependencies between neighboring labels or between a label and the surrounding observations.


See also

Classification, Pattern Recognition, Probability, Sequence

Explore with Wolfram|Alpha

References

Bishop, C. M. Pattern Recognition and Machine Learning. New York: Springer, 2006.

Cite this as:

Weisstein, Eric W. "Sequence Labeling." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/SequenceLabeling.html

Subject classifications