Abstract
In this work, a consensual approach is developed and applied to the noise modeling ofmicrowave transistors. In the proposed method, multiple individual models generated byan expert system ensemble are combined by a consensus rule that results in a consistentand improved generalization outputting with the highest possible reliability andaccuracy. Here the expert system ensemble is basically constructed by the competitorand diverse regressors which in our case are Back-Propagation (BP) Artificial NeuralNetwork (ANN), Support Vector Regression Machine (SVRM), k-Nearest Neighbor (k-NN) and Least Squares (LS) algorithms that perform generalization independently fromeach other.
Keywords
Generalization
Noise (video)
Computer science
Reliability (semiconductor)
Artificial intelligence
k-nearest neighbors algorithm
Support vector machine
Machine learning
Data mining
Algorithm
Mathematics
Physics
Subject Areas
Neural Networks and Applications
·Artificial Intelligence ·Physical Sciences