Background:
Volatile compounds in fish sauce may vary due to the species of fish, ingredients,
processing period, temperature, and even the preference of people in each area. It is necessary
to study a method of distinguishing the origins of fish sauce. The aims of this paper are to introduce a
method to classification of fish sauce origin by means of electronic nose fingerprint and gas chromatography-
mass spectrometry of volatile compounds and the two artificial neural networks are used to
predict the origins of fish sauce.
Methods:
Headspace sampling-solid phase microextraction combined with gas chromatography-mass
spectrometric analysis and electronic nose were used to analysze volatile compounds in different origins
of fish sauce, and these dates predicted the origins of fish sauce by artificial neural networks.
Results:
94 volatile compounds were identified by Automatic mass spectral deconvolution and identification
system, out of which 44 are from Guangdong, 53 from our laboratory, 51 from Vietnam, 47
and 45 from Thailand. Then electronic nose was applied to identify the origin of fish sauce, and the
data were analyzed using principal component analysis and load analysis. The fish sauce from different
origin can be classified well on the PCA plot. Lastly, two artificial neural networks are used to
predict the origins of fish sauce, and the accuracy rates of radial basis and gradient descent both are
93.33%.
Conclusion:
That illustrates that we can provide a quick method to distinguish fish sauce products of
different origins. These results indicated that the combinations of multiple analysis and identification
methods could make up the limitations of a single method, enhance the accuracy of identification, and
provide useful information for product development.
An optimized model for classification of appropriate technology products using neural networks and genetic algorithms
