Click here for full text:
Adaptive Online Learning of Bayesian Network Parameters
Cohen, Ira; Bronstein, Alexandre; Cozman, Fabio G.
HPL-2001-156
Keyword(s): Bayesian networks; machine learning; adaptive systems
Abstract: This abstract contains mathematical formulae which cannot be represented here. The paper introduces Voting EM, an adaptive online learning algorithm of Bayesian network parameters. Voting EM is an extension of the EM (n) algorithm suggested by [1]. We show convergence properties of the Voting EM that uses a constant learning rate. We use the convergence properties to formulate an error driven scheme for adapting the learning rate. The resultant algorithm converges with the optimal rate of 1/t near a maximum while retaining the ability to increase the learning rate in the vicinity of a local maximum or due to changes in the modelled environment.
7 Pages
Back to Index
| 
printable version
