An Artificial Perceptual Model Approach - Artificial Human Sensors: Science and Applications - page 126

The main issue in using an effective methodology to verify the system perfor-
mance is to:
— make related classification of the data,
— observe class behaviours,
— evaluate the analysis process, and
— estimate the accuracy of the presented “environmental picture”.
The make, observe, evaluate and estimate objectives may in a verification
phase of a system's performance be useful tasks to process. The observation ex-
periments are necessary when a direct manipulation of data is not possible. The
observation experiment by the issues: make, observe, evaluate and estimate prin-
ciples, seems to answer important values in getting a credible response. However,
the input situation may of course also affect the trustworthy process in validating
the expected situations.
5.4.1 Human Evaluation
Human evaluation of food products contains a specific description of a regu-
lar, structural and objective procedure. The evaluation in sensory analysis is
a scientific discipline that measures, analyses and interprets responses on food
behaviours that are related to the human perception. This simplified definition
of sensory analysis include both a quantitative as well as a qualitative approach
and may also involve measurements from consumers, trained experts, and cor-
related artificial sensor systems, i.e., perceptual based devices. A strategic com-
bination of these competence's create a balanced knowledge in cooperation, that
in most situations will exhibit a useful and effectively composed evaluation panel
to explore and verify food related products. From the human perspective, it is
interesting to notice that in order to strive in further development of composing
healthy and tasty products, we actually use the human sensing organs as refer-
ences. In understanding the human behaviour when experiencing food, we pro-
vide human panels that use their individual sensing organs to measure properties,
like appearance, smell, taste and consistency for example as illustrated in Fig. 5.16.
The descriptive analysis, Gacula (1997), is a scientific area successfully establishing
quality control to maintain the expected sensory quality parameters in testing of
products in concordance with consumers' response. A useful reference in the used
techniques in sensory evaluations can be found in Meilgaard (2007).
The specific human perception is often also used when estimating the mea-
sured properties from artificial instruments. The calibration and verification of an
artificial perceptual system is complex and the response is qualitatively dependent.
The experimental methodology is often compared and calibrated with human
“experts” or expert panels. Expert in this sense means that the person has proven
perceptual abilities and is able to put words to a sensing impression. Of course,
ordinary consumers also may be involved in the panels.