Information processing apparatus, information processing method, learned model generation method, and program
An information processing apparatus includes: a first acquisition unit that acquires odor data obtained by measuring the odor of an object; a second acquisition unit that acquires an acquisition condition for the odor data; and an identification unit that identifies the object from the odor data and the acquisition condition acquired by the first and second acquisition units on the basis of a learned model obtained by performing learning on the odor data of the object and the acquisition condition, and on the object corresponding to the odor data. The acquisition condition can be text data that indicates the category of the odor and is inputted by a user who has measured the odor.
1. An information processing apparatus, comprising:
a controller configured to execute a program so as to:
acquire target odor data measuring an odor of a target object via an odor sensor using a quartz crystal, the target odor data corresponding to time-series data of frequencies, the frequencies being measured while the odor of the target object is in an inner space of the odor sensor for a predetermined period of time;
acquire a target acquisition condition of the target odor data, the target acquisition condition including environmental information regarding a measurement environment in which the odor of the target object is measured, the environmental information including location information or weather information of the measurement environment;
store data of a plurality of learned models obtained by learning sample odor data and a sample acquisition condition including the location information or the weather information with respect to a plurality of different sample objects, and each of the plurality of learned models having:
an input layer accepting the sample odor data as the time-series data of the frequencies together with the sample acquisition condition in association with the plurality of different sample objects;
an intermediate layer extracting of a plurality of sample features of the sample odor data from the input layer in association with the plurality of different sample objects; and
an output layer providing an identification result based on the plurality of different sample features;
receive an operation input from a user via an operation interface to select a target model of the plurality of learned models;
input the target odor data as the time-series data of the frequencies together with the target acquisition condition into the input layer of the target model;
cause the intermediate layer of the target model to extract a target feature of the target odor data;
cause the output layer of the target model to provide a target identification result based on the target feature and the plurality of sample features of the target model;
identify the target object based on the target identification result; and
display information relating to the identified target object,
wherein the target identification result corresponds to a probability value indicating how likely the target object corresponds to one of the plurality of different sample objects.
2. The information processing apparatus in claim 1 , wherein:
each of the target and sample acquisition conditions includes text data indicating a category of the odor, and the text data is input via the operation interface by a measurement user who measures the odor.
3. The information processing apparatus in claim 1 , wherein:
each of the target and sample acquisition conditions includes state information indicating a state of the odor sensor when measuring the odor.
4. The information processing apparatus in claim 1 , wherein:
the target identification result includes information indicating whether the target object corresponding to the target odor data corresponds to one of the plurality of different sample objects.
5. The information processing apparatus in claim 1 , wherein:
the information processing apparatus is configured to be connected to a terminal of the user, and
the operation input is received from the terminal.
6. An information processing method for causing a processor to execute a program, the information processing method comprising executing on the processor the steps of:
acquiring target odor data measuring an odor of a target object via an odor sensor using a quartz crystal, the target odor data corresponding to time-series data of frequencies, the frequencies being measured while the odor of the target object is in an inner space of the odor sensor for a predetermined period of time;
acquiring a target acquisition condition of the target odor data, the target acquisition condition including environmental information regarding a measurement environment in which the odor of the target object is measured, the environmental information including location information or weather information of the measurement environment;
storing data of a plurality of learned models obtained by learning sample odor data and a sample acquisition condition including the location information or the weather information with respect to a plurality of different sample objects, and each of the plurality of learned models having:
an input layer accepting the sample odor data as the time-series data of the frequencies together with the sample acquisition condition in association with the plurality of different sample objects;
an intermediate layer extracting of a plurality of sample features of the sample odor data from the input layer in association with the plurality of different sample objects; and
an output layer providing an identification result based on the plurality of different sample features;
receiving an operation input from a user via an operation interface to select a target model of the plurality of learned models;
inputting the target odor data as the time-series data of the frequencies together with the target acquisition condition into the input layer of the target model;
causing the intermediate layer of the target model to extract a target feature of the target odor data;
causing the output layer of the target model to provide a target identification result based on the target feature and the plurality of sample features of the target model;
identifying the target object based on the target identification result; and
displaying information relating to the identified target object,
wherein the target identification result corresponds to a probability value indicating how likely the target object corresponds to one of the plurality of different sample objects.
7. The information processing method in claim 6 , wherein:
each of the target and sample acquisition conditions includes text data indicating a category of the odor, and the text data is input via the operation interface by a measurement user who measures the odor.
8. The information processing method in claim 6 , wherein:
each of the target and sample acquisition conditions includes state information indicating a state of the odor sensor when measuring the odor.
9. The information processing method in claim 6 , wherein:
the target identification result includes information indicating whether the target object corresponding to the target odor data corresponds to one of the plurality of different sample objects.
10. A non-transitory computer-readable medium storing a program for causing a computer to execute a process by a processor so as to perform the steps of:
acquiring target odor data measuring an odor of a target object via an odor sensor using a quartz crystal, the target odor data corresponding to time-series data of frequencies, the frequencies being measured while the odor of the target object is in an inner space of the odor sensor for a predetermined period of time;
acquiring a target acquisition condition of the target odor data, the target acquisition condition including environmental information regarding a measurement environment in which the odor of the target object is measured, the environmental information including location information or weather information of the measurement environment;
storing data of a plurality of learned models obtained by learning sample odor data and a sample acquisition condition including the location information or the weather information with respect to a plurality of different sample objects, and each of the plurality of learned models having:
an input layer accepting the sample odor data as the time-series data of the frequencies together with the sample acquisition condition in association with the plurality of different sample objects;
an intermediate layer extracting of a plurality of sample features of the sample odor data from the input layer in association with the plurality of different sample objects; and
an output layer providing an identification result based on the plurality of different sample features;
receiving an operation input from a user via an operation interface to select a target model of the plurality of learned models;
inputting the target odor data as the time-series data of the frequencies together with the target acquisition condition into the input layer of the target model;
causing the intermediate layer of the target model to extract a target feature of the target odor data;
causing the output layer of the target model to provide a target identification result based on the target feature and the plurality of sample features of the target model;
identifying the target object based on the target identification result; and
displaying information relating to the identified target object,
wherein the target identification result corresponds to a probability value indicating how likely the target object corresponds to one of the plurality of different sample objects.
11. The non-transitory computer-readable medium according to claim 10 , wherein:
each of the target and sample acquisition conditions includes text data indicating a category of the odor, and the text data is input via the operation interface by a measurement user who measures the odor.
12. The non-transitory computer-readable medium according to claim 10 , wherein:
each of the target and sample acquisition conditions includes state information indicating a state of the odor sensor when measuring the odor.
13. The non-transitory computer-readable medium according to claim 10 , wherein:
the target identification result includes information indicating whether the target object corresponding to the target odor data corresponds to one of the plurality of different sample objects.