Difference between revisions of "Ristad and Yianilos 1997 Learning String Edit Distance"
Line 13: | Line 13: | ||
== Summary == | == Summary == | ||
− | In this [[Category::paper]] the authors describe | + | In this [[Category::paper]] the authors describe a stochastic model ([[UsesMethod::HMM | HMM]]) for learning a [[AddressesProblem::String Edit Distance]] function, as well as an efficient [[UsesMethod::Expectation_Maximization | EM]] variant for learning edit costs. The authors then present a stochastic solution to [[AddressesProblem::String Classification]], in which the classification is based on the similarity of an observed string, <math>y^v</math>, to an underlying prototype, <math>x^t</math>, from a class, <math>w</math>. The paper describes an efficient algorithm for inducing a joint probability model <math>p(w, y^v|L, \phi)</math> from a corpus, and this model is used to classify new strings. Finally, the described techniques are applied to the problem of learning [[AddressesProblem::Word Pronunciation]] in conversational speech. |
− | and |
Revision as of 12:28, 22 September 2011
Citation
Ristad, E.S. and Yianilos, P.N. Learning string-edit distance. IEEE Transactions on Pattern Analysis and Machine Intelligence, 20: 522--532, 1998.
Online
- Paper web-site by Yianilos.
Summary
In this paper the authors describe a stochastic model (HMM) for learning a String Edit Distance function, as well as an efficient EM variant for learning edit costs. The authors then present a stochastic solution to String Classification, in which the classification is based on the similarity of an observed string, , to an underlying prototype, , from a class, . The paper describes an efficient algorithm for inducing a joint probability model from a corpus, and this model is used to classify new strings. Finally, the described techniques are applied to the problem of learning Word Pronunciation in conversational speech.