Resumo (PT):
Abstract (EN):
In the present work we formulate a general structure for hybrid models. The structure is inspired in the Mixture of Experts (ME) Network developed by-Jacobs and Jordan (1991). The expert modules are extended to formulate knowledge at different levels of sophistication. The resulting hybrid modular network can be given a probabilistic interpretation. In particular the learning algorithm is viewed as a maximum likelihood parameter estimation. The concepts are illustrated by the application to a baker's yeast production process.
Language:
English
Type (Professor's evaluation):
Scientific
No. of pages:
6