Learning internal representations by error propagation
Parallel distributed processing: explorations in the microstructure of cognition, vol. 1
Hi-index | 0.00 |
In this paper, the regression analysis is treated when the output estimate may take more than one value. This is an extension of the usual regression analysis and such cases may happen when the output is affected by some unknown input. The stochastic model used in this paper is the mixture of probabilistic factor analysis model whose identification scheme has been already developed by Tipping and Bishop. We will show the usefulness of our method by a numerical example.