Learning model generation support apparatus, learning model generation support method, and computer-readable recording medium

Patent metadata

Field	Value
Publication number	US-12066417-B2
Application number	US-201817251337-A
Country	US
Kind code	B2
Filing date	Jun 29, 2018
Priority date	Jun 29, 2018
Publication date	Aug 20, 2024
Grant date	Aug 20, 2024

Abstract

A learning model generation support apparatus 10 is an apparatus for supporting generation of a learning model to be utilized in odor detection using an odor sensor that reacts to a plurality of types of odors. The learning model generation support apparatus 10 includes a data acquisition unit 11 that acquires sensor data output by the odor sensor under specific measurement conditions and condition data specifying the measurement conditions, and inputs, as training data, the acquired sensor data and condition data to a machine learning engine 31 that generates the learning model, and a condition setting unit 12 that acquires a predictive accuracy output by the machine learning engine in response to input of the training data, and sets new measurement conditions for when the odor sensor newly outputs sensor data as training data, based on the acquired predictive accuracy.

First claim

The invention claimed is:

1. A learning model generation support apparatus for supporting generation of a learning model to be utilized in odor detection using an odor sensor that reacts to a plurality of types of odors, the apparatus comprising a computer containing at least one processor, wherein the at least one processor is configured to function as:
a data acquisition unit configured to acquire sensor data output by the odor sensor under a specific measurement condition and condition data specifying the measurement condition, and input the acquired sensor data and the condition data, as training data, to a machine learning engine for generating the learning model;
a condition setting unit configured to acquire a predictive accuracy output by the machine learning engine in response to input of the training data, and based on the predictive accuracy, set a new measurement condition for when the odor sensor newly outputs sensor data as the training data;
wherein the condition setting unit sets the new measurement condition, by executing sequential model-based optimization with a predictive accuracy higher than the acquired predictive accuracy and the measurement condition as parameters, wherein repeating selection of a combination of candidate parameters and executing of simulations using the selected combination of the candidate parameters, changing the combination of the candidate parameters each time a simulation is executed, and specifying the combination of the candidate parameters whose predictive accuracy is higher than the acquired predictive accuracy as the new measurement condition.

2. The learning model generation support apparatus according to claim 1 , wherein
the measurement condition includes at least temperature and humidity ambient to the odor sensor.

3. The learning model generation support apparatus according to claim 2 , wherein
the measurement condition further includes information relating to an odor ambient to the odor sensor, other than an odor to be detected.

4. The learning model generation support apparatus according to claim 1 , further comprising:
a presentation unit configured to present the set measurement condition.

5. The learning model generation support apparatus according to claim 1 , wherein the new measurement conditions are set by setting a search space of parameters divided into a grid, with the number of parameters to be searched, allocating a combination of the parameters for every grid point, executing simulation for every combination, specifying the combination whose predictive accuracy is higher than the acquired predictive accuracy, and setting the specified combination as the new measurement conditions.

6. A learning model generation support method for supporting generation of a learning model to be utilized in odor detection using an odor sensor that reacts to a plurality of types of odors, the
method executed by a computer and comprising:
acquiring sensor data output by the odor sensor under a specific measurement condition and the condition data specifying the measurement condition, and inputting the acquired sensor data and condition data, as training data, to a machine learning engine for generating the learning model;
acquiring a predictive accuracy output by the machine learning engine in response to input of the training data, and based on the predictive accuracy, setting a new measurement condition for when the odor sensor newly outputs sensor data as the training data;
wherein in the setting, the new measurement condition is set, by executing sequential model-based optimization with a predictive accuracy higher than the acquired predictive accuracy and the measurement condition as parameters, wherein repeating selection of a combination of candidate parameters and executing of simulations using the selected combination of the candidate parameters, changing the combination of the candidate parameters each time a simulation is executed, and specifying the combination of the candidate parameters whose predictive accuracy is higher than the acquired predictive accuracy as the new measurement condition.

7. The learning model generation support method according to claim 6 , wherein
the measurement condition includes at least temperature and humidity ambient to the odor sensor.

8. The learning model generation support method according to claim 7 , wherein
the measurement condition further includes information relating to an odor ambient to the odor sensor, other than an odor to be detected.

9. The learning model generation support method according to claim 6 , further comprising:
presenting the set measurement condition.

10. The learning model generation support method according to claim 6 , wherein the new measurement conditions are set by setting a search space of parameters divided into a grid, with the number of parameters to be searched, allocating a combination of the parameters for every grid point, executing simulation for every combination, specifying the combination whose predictive accuracy is higher than the acquired predictive accuracy, and setting the specified combination as the new measurement conditions.

11. A non-transitory computer-readable recording medium that includes a program recorded thereon for supporting generation of a learning model to be utilized in odor detection using an odor sensor that reacts to a plurality of types of odors, with a computer, the program including instructions that cause the computer to carry out:
acquiring sensor data output by the odor sensor under a specific measurement condition and the condition data specifying the measurement condition, and inputting the acquired sensor data and condition data, as training data, to a machine learning engine for generating the learning model;
acquiring a predictive accuracy output by the machine learning engine in response to input of the training data, and based on the predictive accuracy, setting a new measurement condition for when the odor sensor newly outputs sensor data as the training data;
wherein in the setting, the new measurement condition is set, by executing sequential model-based optimization with a predictive accuracy higher than the acquired predictive accuracy and the measurement condition as parameters, wherein repeating selection of a combination of candidate parameters and executing of simulations using the selected combination of the candidate parameters, changing the combination of the candidate parameters each time a simulation is executed, and specifying the combination of the candidate parameters whose predictive accuracy is higher than the acquired predictive accuracy as the new measurement condition.

12. The non-transitory computer-readable recording medium according to claim 11 , wherein
the measurement condition includes at least temperature and humidity ambient to the odor sensor.

13. The non-transitory computer-readable recording medium according to claim 12 , wherein
the measurement condition further includes information relating to an odor ambient to the odor sensor, other than an odor to be detected.

14. The non-transitory computer-readable recording medium according to claim 11 , wherein
the program further includes instructions that cause the computer to carry out:
a step of presenting the set measurement condition.

15. The non-transitory computer-readable recording medium according to claim 11 , wherein the new measurement conditions are set by setting a search space of parameters divided into a grid, with the number of parameters to be searched, allocating a combination of the parameters for every grid point, execu

Assignees

• Nec Corp

Inventors

• Eto Riki

Classifications

• G05B13/048

  using a predictor · CPC title

• G05B13/0265

  the criterion being a learning criterion · CPC title

• G06N20/00

  Machine learning · CPC title

• G01N19/00

  Investigating materials by mechanical methods (G01N3/00 - G01N17/00 take precedence) · CPC title

• G01N33/0034 Primary

  comprising neural networks or related mathematical techniques · CPC title