Reduced false positive identification for spectroscopic quantification
A device may receive information identifying results of a spectroscopic measurement performed on an unknown sample. The device may determine a decision boundary for a quantification model based on a configurable parameter, such that a first plurality of training set samples of the quantification model is within the decision boundary and a second plurality of training set samples of the quantification model is not within the decision boundary. The device may determine a distance metric for the spectroscopic measurement performed on the unknown sample relative to the decision boundary. The device may determine a plurality of distance metrics for the second plurality of training set samples of the quantification model relative to the decision boundary. The device may provide information indicating whether the spectroscopic measurement performed on the unknown sample corresponds to the quantification model.
1 . A method, comprising:
generating, by a control device and using a machine learning technique, a single class representing a quantification model by aggregating multiple classes corresponding to multiple concentrations of a component in a training set; and
one or more of:
using the single class representing the quantification model to perform spectroscopic quantification of an unknown sample, or
providing, to one or more other control devices and for utilization in spectroscopic quantification of other unknown samples, the single class representing the quantification model.
2 . The method of claim 1 ,
wherein the machine learning technique is a single class support vector machine (SC-SVM) technique, and
wherein the method further comprises:
selecting to use the SC-SVM technique for multiple quantification models that include the quantification model.
3 . The method of claim 1 , wherein the quantification model includes information associated with assigning a particular spectrum to a particular concentration of the multiple concentrations.
4 . The method of claim 1 ,
wherein the method comprises using the single class representing the quantification model, and
wherein using the single class representing the quantification model comprises:
determining a configurable parameter value for using the machine learning technique;
determining, based on the configurable parameter value, whether the unknown sample corresponds to the quantification model; and
providing information indicating whether the unknown sample corresponds to the quantification model.
5 . The method of claim 4 , wherein the configurable parameter value corresponds to a ratio of training set samples that are determined to be within a decision boundary for the machine learning technique to training set samples that are determined to not be within the decision boundary.
6 . The method of claim 4 , wherein determining whether the unknown sample corresponds to the quantification model comprises:
determining a decision boundary based on the configurable parameter value; and
determining, based on the decision boundary, whether the unknown sample corresponds to the quantification model.
7 . The method of claim 4 , wherein determining whether the unknown sample corresponds to the quantification model comprises:
setting a decision boundary to satisfy the configurable parameter value;
determining a location on a set of axes for the unknown sample;
determining a distance between the location and a nearest point of the decision boundary; and
determining, based on the distance, whether the unknown sample corresponds to the quantification model.
8 . The method of claim 4 , wherein determining whether the unknown sample corresponds to the quantification model comprises:
determining a decision boundary based on the configurable parameter value; and
determining that the unknown sample corresponds to the quantification model based on a measurement being less than a threshold quantity of standard deviations from the decision boundary.
9 . The method of claim 1 , further comprising:
determining that the unknown sample corresponds to the quantification model; and
performing, based on determining that the unknown sample corresponds to the quantification model, one or more spectroscopic determinations based on a result of the spectroscopic quantification.
10 . A control device, comprising:
one or more memories; and
one or more processors, coupled to the one or more memories, configured to:
generate, using a machine learning technique, a single class representing a quantification model by aggregating multiple classes corresponding to multiple concentrations of a component in a training set; and
one or more of:
use the single class representing the quantification model to perform spectroscopic quantification of an unknown sample, or
provide, to one or more other control devices and for utilization in spectroscopic quantification of other unknown samples, the single class representing the quantification model.
11 . The control device of claim 10 ,
wherein the machine learning technique is a single class support vector machine (SC-SVM) technique, and
wherein the one or more processors are further configured to:
select to use the SC-SVM technique for multiple quantification models that include the quantification model.
12 . The control device of claim 10 , wherein the quantification model includes information associated with assigning a particular spectrum to a particular concentration of the multiple concentrations.
13 . The control device of claim 10 ,
wherein the one or more processors are configured to use the single class representing the quantification model, and
wherein the one or more processors, to use the single class representing the quantification model, are configured to:
determine a configurable parameter value for using the machine learning technique;
determine, based on the configurable parameter value, whether the unknown sample corresponds to the quantification model; and
provide information indicating whether the unknown sample corresponds to the quantification model.
14 . The control device of claim 13 , wherein the configurable parameter value corresponds to a ratio of training set samples that are determined to be within a decision boundary for the machine learning technique to training set samples that are determined to not be within the decision boundary.
15 . The control device of claim 13 , wherein the one or more processors, to determine whether the unknown sample corresponds to the quantification model, are configured to:
determine a decision boundary based on the configurable parameter value; and
determine, based on the decision boundary, whether the unknown sample corresponds to the quantification model.
16 . The control device of claim 13 , wherein the one or more processors, to determine whether the unknown sample corresponds to the quantification model, are configured to:
set a decision boundary to satisfy the configurable parameter value;
determine a location on a set of axes for the unknown sample;
determine a distance between the location and a nearest point of the decision boundary; and
determine, based on the distance, whether the unknown sample corresponds to the quantification model.
17 . The control device of claim 13 , wherein the one or more processors, to determine whether the unknown sample corresponds to the quantification model, are configured to:
determine a decision boundary based on the configurable parameter value; and
determine that the unknown sample corresponds to the quantification model based on a measurement being less than a threshold quantity of standard deviations from the decision boundary.
18 . The control device of claim 10 , wherein the one or more processors are further configured to:
determine that the unknown sample corresponds to the quantification model; and
perform, based on determining that the unknown sample corresponds to the quantification model, one or more spectroscopic determinations based on a result of the spectroscopic quantification.
19 . A non-transitory computer-readable medium storing a set of instructions, the set of instructions comprising:
one or more instructions that, when executed by one or more processors of a control device, cause the control device to:
generate, using a machine learning technique, a single class representing a quantification model by aggregating multiple classes corresponding to multiple concentrations of a component in a training set; and
one or more of:
use the single class representing the quantification model to perform spectroscopic quantification of an unknown sample, or
provide, to one or more other control devices and for utilization in spectroscopic quantification of other unknown samples, the single class representing the quantification model.
20 . The non-transitory computer-readable medium of claim 19 , wherein the one or more instructions further cause the control device to:
select to use the machine learning technique for multiple quantification models that include the quantification model.