Limits...
Semi-supervised Bayesian classification of materials with impact-echo signals.

Igual J, Salazar A, Safont G, Vergara L - Sensors (Basel) (2015)

Bottom Line: This spectrum is the input data to a Bayesian classifier that is based on the modeling of the conditional probabilities with a mixture of Gaussians.The advantage of our proposal is that it is flexible, since it obtains good results for a wide range of models even under little supervision; e.g., it obtains a harmonic average of precision and recall value of 92.38% given only a 10% supervision ratio.We test the method with real specimens made of aluminum alloy.

View Article: PubMed Central - PubMed

Affiliation: Departamento de Comunicaciones, Universitat Politecnica de Valencia, Camino de Vera s/n, 46022 Valencia, Spain. jigual@dcom.upv.es.

ABSTRACT
The detection and identification of internal defects in a material require the use of some technology that translates the hidden interior damages into observable signals with different signature-defect correspondences. We apply impact-echo techniques for this purpose. The materials are classified according to their defective status (homogeneous, one defect or multiple defects) and kind of defect (hole or crack, passing through or not). Every specimen is impacted by a hammer, and the spectrum of the propagated wave is recorded. This spectrum is the input data to a Bayesian classifier that is based on the modeling of the conditional probabilities with a mixture of Gaussians. The parameters of the Gaussian mixtures and the class probabilities are estimated using an extended expectation-maximization algorithm. The advantage of our proposal is that it is flexible, since it obtains good results for a wide range of models even under little supervision; e.g., it obtains a harmonic average of precision and recall value of 92.38% given only a 10% supervision ratio. We test the method with real specimens made of aluminum alloy. The results show that the algorithm works very well. This technique could be applied in many industrial problems, such as the optimization of the marble cutting process.

No MeSH data available.


Related in: MedlinePlus

Dimension reduction of the feature vector after PCA. Box and whiskers plot of the F measure for each of the 12 classes. (Top) 3 × 1 feature vector; (Middle) 7 × 1 feature vector; (Bottom) 16 × 1 feature vector. Above each plot is the overall mean F measure.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4481956&req=5

f4-sensors-15-11528: Dimension reduction of the feature vector after PCA. Box and whiskers plot of the F measure for each of the 12 classes. (Top) 3 × 1 feature vector; (Middle) 7 × 1 feature vector; (Bottom) 16 × 1 feature vector. Above each plot is the overall mean F measure.

Mentions: The first thing we have to establish is the dimension of the feature space, i.e., the number of principal components that we are going to keep after PCA. We will use the 12-class problem to determine the feature space dimension, since it is the most complicated case. To analyze the influence of the dimension of the feature vector, we run the algorithm for different dimensions. We obtain the confusion matrix and then calculate the F value. In Figure 4, we show the box and whiskers plot of the F value for all of the classes and the overall F mean value (40 runs for each one) when the feature vector is a 3 × 1 vector (top), a 7 × 1 vector (middle) and a 16 × 1 vector (bottom). These values correspond to keeping 25%, 50% and 75% of the total amount of the variance when applying PCA. The overall F mean values are: 83.09, 92.38 and 88.46, respectively.


Semi-supervised Bayesian classification of materials with impact-echo signals.

Igual J, Salazar A, Safont G, Vergara L - Sensors (Basel) (2015)

Dimension reduction of the feature vector after PCA. Box and whiskers plot of the F measure for each of the 12 classes. (Top) 3 × 1 feature vector; (Middle) 7 × 1 feature vector; (Bottom) 16 × 1 feature vector. Above each plot is the overall mean F measure.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4481956&req=5

f4-sensors-15-11528: Dimension reduction of the feature vector after PCA. Box and whiskers plot of the F measure for each of the 12 classes. (Top) 3 × 1 feature vector; (Middle) 7 × 1 feature vector; (Bottom) 16 × 1 feature vector. Above each plot is the overall mean F measure.
Mentions: The first thing we have to establish is the dimension of the feature space, i.e., the number of principal components that we are going to keep after PCA. We will use the 12-class problem to determine the feature space dimension, since it is the most complicated case. To analyze the influence of the dimension of the feature vector, we run the algorithm for different dimensions. We obtain the confusion matrix and then calculate the F value. In Figure 4, we show the box and whiskers plot of the F value for all of the classes and the overall F mean value (40 runs for each one) when the feature vector is a 3 × 1 vector (top), a 7 × 1 vector (middle) and a 16 × 1 vector (bottom). These values correspond to keeping 25%, 50% and 75% of the total amount of the variance when applying PCA. The overall F mean values are: 83.09, 92.38 and 88.46, respectively.

Bottom Line: This spectrum is the input data to a Bayesian classifier that is based on the modeling of the conditional probabilities with a mixture of Gaussians.The advantage of our proposal is that it is flexible, since it obtains good results for a wide range of models even under little supervision; e.g., it obtains a harmonic average of precision and recall value of 92.38% given only a 10% supervision ratio.We test the method with real specimens made of aluminum alloy.

View Article: PubMed Central - PubMed

Affiliation: Departamento de Comunicaciones, Universitat Politecnica de Valencia, Camino de Vera s/n, 46022 Valencia, Spain. jigual@dcom.upv.es.

ABSTRACT
The detection and identification of internal defects in a material require the use of some technology that translates the hidden interior damages into observable signals with different signature-defect correspondences. We apply impact-echo techniques for this purpose. The materials are classified according to their defective status (homogeneous, one defect or multiple defects) and kind of defect (hole or crack, passing through or not). Every specimen is impacted by a hammer, and the spectrum of the propagated wave is recorded. This spectrum is the input data to a Bayesian classifier that is based on the modeling of the conditional probabilities with a mixture of Gaussians. The parameters of the Gaussian mixtures and the class probabilities are estimated using an extended expectation-maximization algorithm. The advantage of our proposal is that it is flexible, since it obtains good results for a wide range of models even under little supervision; e.g., it obtains a harmonic average of precision and recall value of 92.38% given only a 10% supervision ratio. We test the method with real specimens made of aluminum alloy. The results show that the algorithm works very well. This technique could be applied in many industrial problems, such as the optimization of the marble cutting process.

No MeSH data available.


Related in: MedlinePlus