Collective Neural Classifiers for Food Quality Applications

Inhalt:
The paper studies the possibility for increasing neural networks classification accuracy when applying pattern recognition methods to food quality determination. More specifically, the simulation study used training set of spectral patterns of potatoes expertly classified into three quality classes and neural networks for determination egg freshness. The basis classifier in the products evaluation problem was a neural classifier, multi-layer perceptron (MLP). Classification accuracy was boosted by enhancing the synthesis of collective decision rules, the AdaBoost method in particular.