Table 1.
Specifications of the involved sensors of the custom e-nose.
Table 2.
Description of the two custom bacteria odor datasets.
Table 3.
Previously literatures on feature evaluation for relevance based on linear transformation.
Table 4.
Algorithm flow of the IFE method.
Fig 1.
Schematic diagram of FGS.
Fig 2.
Random points from norm distribution with different lambda.
Table 5.
Structure description of the eight datasets.
Fig 3.
Overview diagram of the IFE-FGE on the bacteria datasets.
Table 6.
Initial performance of the e-nose with the use of all sensors and single feature extraction method.
Table 7.
Experiment results of all the involved algorithms on the two bacteria datasets and six famous public datasets.
Table 8.
The experimental results (Mean) of all the involved algorithms on six public datasets.
Fig 4.
The experimental results of all the involved algorithms on the bacteria datasets 1.
Fig 5.
The experimental results of all the involved algorithms on the bacteria datasets 2.
Fig 6.
Heatmap matrix (Mean accuracy) for rank statistics of the validation results on the eight datasets.
Fig 7.
Heatmap matrix (Max accuracy) for rank statistics of the validation results on the eight datasets.
*The cell denotes the number of corresponding rank (each column) achieved by the model (each row) on the results of the eight datasets.
Table 9.
The experimental results (Max) of all the involved algorithms on six public datasets.
Table 10.
Characteristics of the sensors recommended by IFE-FGS under limit of eight features.
Table 11.
Known volatile organic compounds and gases emanated from the metabolite of the bacteria.