Improved recognition of E. coli 0157:H7 bacteria from pulsed-field gel electrophoresis images through the fusion of neural networks with fuzzy training data

E. coli 0157:H7 is a particularly toxic strain of the E. coli bacteria group. Fast and accurate methodologies for recognizing this strain of bacteria are desired. In this paper, we demonstrate that the fusion of neural network outputs using (possibly redundant) banding pattern features from pulsed-field electrophoresis images, where the learning was done with fuzzy labeled training data, produced superior results over the single and crisp counterparts.