Smell estimation device, smell estimation system, smell estimation method, and smell estimation program
The odor estimation device improves accuracy by using a trained model to estimate subjective odor data and retraining when discrepancies are detected, addressing inconsistencies between AI and human perception.
Patent Information
- Application Number
- JP2023076934
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-08
- Publication Date
- 2024-11-20
AI Technical Summary
Existing odor estimation technologies using AI face inaccuracies due to differences between subjective odor data estimated by AI and humans, and fail to correctly detect changes in odor over time, leading to inconsistent estimation results.
An odor estimation device and system that utilizes a trained model to estimate subjective odor data from sensor data, includes an estimation unit to obtain estimated subjective data, a learning unit to retrain the model when discrepancies are detected, and a notification unit to prompt user input for improving accuracy.
Reduces the difference between AI-estimated and human subjective odor data by continuously retraining the model based on user input, enhancing the accuracy of odor estimation.
Smart Images

Figure 2024161842000001_ABST
Abstract
Description
[Technical field]
[0001] The present disclosure relates to an odor estimation device, an odor estimation system, an odor estimation method, and an odor estimation program. [Background technology]
[0002] Technologies have been disclosed that focus on the difference between the output of an odor sensor that detects an odor and the sense of a human who smells the same odor. Patent Document 1 discloses a technology that acquires data from an odor sensor and subjective data about odors by humans and uses AI (Artificial Intelligence) to learn the data. Patent Document 2 discloses a technology that allows a user to input data about odors detected by an odor sensor and whether they like or dislike the odor. [Prior art documents] [Patent documents]
[0003] [Patent Document 1] International Publication No. 2022 / 176923 [Patent Document 2] International Publication No. 2021 / 171987 Summary of the Invention [Problem to be solved by the invention]
[0004] However, there may be a difference between the subjective data of the smell estimated by AI and the subjective data of the smell judged by a human. In Patent Document 1, machine learning is performed using a set of the evaluation for each index and the smell information detected by the smell sensor as training data, but if the difference is left unchecked, the estimation result by the AI may be inaccurate, and the technology disclosed in Patent Document 1 may not be able to correctly detect the change in the smell in the space where the smell sensor is installed. In addition, the technology disclosed in Patent Document 2 has the user input whether they like or dislike the smell, and does not obtain the difference between the subjective data of the smell estimated by AI and the subjective data of the smell judged by a human.
[0005] The present disclosure has been made in consideration of the above-mentioned points, and aims to provide an odor estimation device, odor estimation system, odor estimation method, and odor estimation program that reduce the difference between subjective data of odors estimated by AI and subjective data of odors estimated by humans. [Means for solving the problem]
[0006] The odor estimation device according to the first aspect of the present disclosure comprises an estimation unit that uses a trained model to estimate subjective data of an odor from sensor data that continuously senses an odor and is continuously transmitted from an odor sensor via a terminal, and a learning unit that re-trains the trained model so that, if there is a difference between the estimated subjective data and the input subjective data at the evaluation date and time in input subjective data that includes information on the subjectivity of the odor and the evaluation date and time when the odor was evaluated and that is input to the terminal by a user who smelled the odor, outputs the input subjective data by inputting the sensor data corresponding to the odor.
[0007] The odor estimation device according to a second aspect of the present disclosure is the odor estimation device according to the first aspect, further comprising a notification unit that notifies the terminal to input the input subjective data.
[0008] An odor estimation device according to a third aspect of the present disclosure is the odor estimation device according to the second aspect, wherein the notification unit notifies the terminal to input the input subjective data when the accuracy of the estimated subjective data output by the trained model is equal to or lower than a predetermined threshold.
[0009] An odor estimation device according to a fourth aspect of the present disclosure is the odor estimation device according to the third aspect, wherein the notification unit notifies the terminal to input the input subjective data when the accuracy of the estimated subjective data output by the trained model remains below a predetermined threshold for a predetermined period of time.
[0010] An odor inference device according to a fifth aspect of the present disclosure is the odor inference device according to the first aspect, wherein the sensor data includes information on the time at which the odor sensor performed sensing.
[0011] An odor estimation system according to a sixth aspect of the present disclosure comprises an odor sensor that continuously senses an odor to obtain sensor data, a terminal connected to the odor sensor to obtain the sensor data, and an odor estimation device that obtains the sensor data from the terminal via wireless communication, wherein the odor estimation device comprises an estimation unit that uses a trained model that estimates subjective data of an odor from the sensor data of continuously sensed odors that is continuously transmitted from the odor sensor through the terminal, and a learning unit that re-trains the trained model to output the input subjective data by inputting the sensor data corresponding to the odor if there is a difference between the estimated subjective data and the input subjective data at the evaluation date and time input to the terminal by a user who smelled the odor, the input subjective data including information on the subjectivity of the odor and the evaluation date and time when the odor was evaluated.
[0012] An odor estimation method according to a seventh aspect of the present disclosure includes a processor that uses a trained model to estimate subjective data of an odor from sensor data that continuously senses an odor and is continuously transmitted from an odor sensor through a terminal, and if there is a difference between the estimated subjective data and the input subjective data at the evaluation date and time input to the terminal by a user who smelled the odor, the input subjective data including information on the subjectiveness of the odor and the evaluation date and time when the odor was evaluated, performs a process of re-training the trained model to output the input subjective data by inputting the sensor data corresponding to the odor.
[0013] An odor estimation program according to an eighth aspect of the present disclosure causes a computer to obtain estimated subjective data of an odor using a trained model that estimates subjective data of an odor from sensor data that continuously senses an odor and is continuously transmitted from an odor sensor through a terminal, and if there is a difference between the estimated subjective data and the input subjective data at the evaluation date and time input to the terminal by a user who smelled the odor, which includes information on the subjectiveness of the odor and the evaluation date and time when the odor was evaluated, execute a process of re-training the trained model so as to output the input subjective data by inputting the sensor data corresponding to the odor. Effect of the Invention
[0014] According to the present disclosure, it is possible to provide an odor estimation device, an odor estimation system, an odor estimation method, and an odor estimation program that reduce the difference between subjective data of an odor estimated by AI and subjective data of an odor estimated by humans. [Brief description of the drawings]
[0015] [Figure 1] 1 is a diagram showing a schematic configuration of an odor detection system including an information processing device according to an embodiment of the disclosed technology. [Diagram 2] FIG. 2 is a block diagram showing a hardware configuration of the information processing device. [Diagram 3] FIG. 2 is a block diagram showing a hardware configuration of a mobile terminal. [Figure 4] 2 is a block diagram showing an example of a functional configuration of an information processing device and a mobile terminal; [Diagram 5] 13 is a flowchart showing the flow of an odor estimation process performed by an information processing device. [Figure 6A] FIG. 13 is a diagram showing an example of a keyword input screen presented by the mobile terminal when inputting subjective data. [Figure 6B] FIG. 13 is a diagram showing an example of a keyword input screen presented by the mobile terminal when inputting subjective data. [Figure 6C]FIG. 13 is a diagram showing an example of a keyword input screen presented by the mobile terminal when inputting subjective data. [Figure 6D] FIG. 13 is a diagram showing an example of a keyword input screen presented by the mobile terminal when inputting subjective data. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0016] Hereinafter, an example of an embodiment of the present disclosure will be described with reference to the drawings. In each drawing, the same or equivalent components and parts are given the same reference numerals. In addition, the dimensional ratios of the drawings are exaggerated for the convenience of explanation and may differ from the actual ratios.
[0017] 1 is a diagram showing a schematic configuration of an odor detection system including an information processing device according to the present embodiment. The odor detection system according to the present embodiment includes an information processing device 10, an odor sensor 20, a mobile terminal 30, and a display device 40.
[0018] The information processing device 10 is an example of an odor estimation device of the present disclosure, and is a device that estimates what an odor sensed by the odor sensor 20 smells like from a human subjective perspective. The information processing device 10 uses a trained model that has been generated in advance by machine learning in order to estimate what an odor sensed by the odor sensor 20 smells like from a human subjective perspective. The information processing device 10 is a device that is installed in a location different from where the odor sensor 20 is installed, and may be, for example, a cloud server.
[0019] The odor sensor 20 is a sensor that constantly senses odors in the space in which it is installed. In other words, the odor sensor 20 is not provided for sensing odors released in one shot, but is provided for constantly sensing odors in the space in which it is installed. The odor sensor 20 is, for example, a membrane-type surface stress sensor that outputs a signal in response to the adhesion and detachment of molecules that constitute odor components, and the odor sensor 20 outputs an electrical signal having a voltage waveform in response to the adhesion and detachment of the molecules as a sensing result. The odor sensor 20 may be, for example, a membrane-type surface stress sensor, or various types of sensors such as a cantilever sensor, an optical sensor, a piezoelectric sensor, or a vibration response sensor.
[0020] The odor sensor 20 is connected to a mobile terminal 30, and the sensing results of the odor sensor 20 are sent to the mobile terminal 30. The mobile terminal 30 constantly transmits the sensing results of the odor sensor 20 as sensor data to the information processing device 10 via wireless communication. The information processing device 10 then acquires the sensor data sent from the mobile terminal 30, and constantly estimates what the odor sensed by the odor sensor 20 smells like from a human subjective perspective. The estimation result by the information processing device 10 is regarded as estimated subjective data. The information processing device 10 outputs the estimated subjective data to the display device 40 or the mobile terminal 30.
[0021] The mobile terminal 30 is a device such as a smartphone, and is connected to the odor sensor 20 by wire. The mobile terminal 30 has a function of constantly transmitting the sensing results of the odor sensor 20 to the information processing device 10, as well as a function of allowing the user to input what kind of odor the space in which the odor sensor 20 is placed is, that is, what kind of odor the user considers to be. The odor sensor 20 regards the data on what kind of odor the user has inputted as input subjective data.
[0022] Here, the input subjective data input by the user does not always match the estimated subjective data estimated by the information processing device 10, and there are cases where the input subjective data differs from the estimated subjective data. When the input subjective data differs from the estimated subjective data, the information processing device 10 re-trains the trained model so as to output the input subjective data in response to input of sensor data corresponding to the same odor.
[0023] When the input subjective data and the estimated subjective data differ, the information processing device 10 can improve the accuracy of odor estimation compared to a case in which the trained model is not retrained by retraining the trained model so that the input subjective data is output in response to input of sensor data corresponding to the same odor.
[0024] FIG. 2 is a block diagram showing a hardware configuration of the information processing device 10. As shown in FIG.
[0025] 2, the information processing device 10 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a storage 14, an input unit 15, a display unit 16, and a communication interface (I / F) 17. Each component is connected to each other via a bus 19 so as to be able to communicate with each other.
[0026] The CPU 11 is a central processing unit that executes various programs and controls each part. That is, the CPU 11 reads a program from the ROM 12 or the storage 14, and executes the program using the RAM 13 as a working area. The CPU 11 controls each of the above components and performs various arithmetic processing according to the program recorded in the ROM 12 or the storage 14. In this embodiment, the ROM 12 or the storage 14 stores an odor estimation program that estimates what kind of odor the odor sensed by the odor sensor 20 smells like to a human subjective sense.
[0027] The ROM 12 stores various programs and various data. The RAM 13 temporarily stores programs or data as a working area. The storage 14 is configured with a storage device such as a hard disk drive (HDD), a solid state drive (SSD), or a flash memory, and stores various programs including an operating system, and various data.
[0028] The input unit 15 includes a pointing device such as a mouse, and a keyboard, and is used to perform various input operations.
[0029] The display unit 16 is, for example, a liquid crystal display, and displays various information. The display unit 16 may function as the input unit 15 by adopting a touch panel system.
[0030] The communication interface 17 is an interface for communicating with other devices such as the mobile terminal 30, and uses standards such as Ethernet (registered trademark), FDDI, and Wi-Fi (registered trademark), for example.
[0031] FIG. 3 is a block diagram showing the hardware configuration of the mobile terminal 30. As shown in FIG.
[0032] 3, the mobile terminal 30 includes a CPU 31, a ROM 32, a RAM 33, a storage 34, an input unit 35, a display unit 36, and a communication interface (I / F) 37. Each component is connected via a bus 39 so as to be able to communicate with each other.
[0033] The CPU 31 is a central processing unit that executes various programs and controls each part. That is, the CPU 31 reads a program from the ROM 32 or the storage 34, and executes the program using the RAM 33 as a working area. The CPU 31 controls each of the above components and performs various arithmetic processing according to the program recorded in the ROM 32 or the storage 34. In this embodiment, the ROM 32 or the storage 34 stores a data processing program that performs processing on the sensor data acquired from the odor sensor 20.
[0034] The ROM 32 stores various programs and various data. The RAM 33 temporarily stores programs or data as a working area. The storage 34 is configured with a storage device such as a flash memory, and stores various programs including an operating system, and various data.
[0035] The input unit 35 is used to perform various input operations.
[0036] The display unit 36 is, for example, a liquid crystal display, and displays various information. The display unit 36 may function as the input unit 35 by adopting a touch panel system.
[0037] The communication interface 37 is an interface for communicating with other devices such as the information processing device 10, and for this communication, a wireless communication standard such as 4G, 5G, or Wi-Fi (registered trademark) is used.
[0038] When executing the odor estimation program, the information processing device 10 realizes various functions using the hardware resources shown in Fig. 2. When executing the data processing program, the mobile terminal 30 realizes various functions using the hardware resources shown in Fig. 3. The functional configurations realized by the information processing device 10 and the mobile terminal 30 will be described.
[0039] FIG. 4 is a block diagram showing an example of a functional configuration of the information processing device 10 and the mobile terminal 30. As shown in FIG.
[0040] 4, the information processing device 10 has, as its functional components, data receiving units 101 and 102, an AI learning unit 103, a trained model 104, an AI estimation unit 105, a notification unit 106, and a database 110. Each functional component is realized by the CPU 11 reading out and executing an odor estimation program stored in the ROM 12 or the storage 14.
[0041] 4, the mobile terminal 30 has a continuous data acquisition unit 301, data transmission units 302 and 304, a keyword input unit 303, a time assignment unit 305, and an estimation utilization unit 306. Each functional configuration is realized by the CPU 31 reading and executing a data processing program stored in the ROM 32 or the storage 34.
[0042] First, the functional configuration of the information processing device 10 will be described. The data receiving unit 101 receives sensor data sensed by the odor sensor 20 and transmitted via the mobile terminal 30. The sensor data received by the data receiving unit 101 is stored in a sensor database 111 in a database 110 for storing sensor data. Information on the date and time when the sensor data was acquired by the mobile terminal 30 is added to the sensor data received by the data receiving unit 101. Furthermore, information identifying the odor sensor 20 or information identifying the location where the odor sensor 20 is installed may be added to the sensor data received by the data receiving unit 101.
[0043] The data receiving unit 102 receives input subjective data that is input to the mobile terminal 30 by a user about an odor sensed by the user in the space in which the odor sensor 20 is placed. The input subjective data received by the data receiving unit 102 is stored in a subjective database 112 for storing input subjective data in the database 110. The input subjective data received by the data receiving unit 101 includes information on the evaluation date and time when the user subjectively evaluated the odor.
[0044] The AI learning unit 103 generates a trained model 104 by machine learning using an arbitrary number of sensor data stored in the sensor database 111. The trained model 104 is a trained model for obtaining estimated subjective data from the sensor data received by the data receiving unit 101. The trained model 104 can be generated depending on the type of the odor sensor 20.
[0045] Furthermore, if the estimated subjective data obtained using the trained model 104 based on the sensor data at the evaluation date and time included in the input subjective data differs from the input subjective data, the AI learning unit 103 re-trains the trained model 104 so as to obtain the input subjective data from the sensor data at the evaluation date and time. Re-training the trained model 104 means adjusting the parameters of the trained model 104 so as to output the same subjective data as the input subjective data at the same evaluation date and time using the same sensor data as the sensor data at the evaluation date and time. If there is a difference between the estimated subjective data and the input subjective data, the AI learning unit 103 re-trains the trained model 104, thereby improving the estimation accuracy of the estimated subjective data compared to the case where re-training is not performed.
[0046] The AI estimation unit 105 obtains estimated subjective data from the sensor data using the trained model 104 generated by machine learning in the AI learning unit 103. Obtaining estimated subjective data using the trained model 104 means inputting sensor data acquired from the mobile terminal 30 to the trained model 104 and outputting estimated subjective data from the trained model 104. The estimated subjective data obtained by the AI estimation unit 105 is displayed, for example, on the display unit 401 of the display device 40. The estimated subjective data is displayed, for example, on a Web browser. The estimated subjective data obtained by the AI estimation unit 105 is sent, for example, to the mobile terminal 30. The trained model 104 used by the AI estimation unit 105 to obtain estimated subjective data may be generated by machine learning so as to output, in addition to the estimated subjective data, information on the accuracy of the estimated subjective data.
[0047] The estimated subjective data estimated by the AI estimation unit 105 using the trained model 104 corresponds to keywords used for inputting subjective data in the mobile terminal 30 described later. Therefore, the trained model 104 is machine-learned in advance so as to output keywords used for inputting subjective data in the mobile terminal 30 from the sensor data.
[0048] The notification unit 106 notifies the mobile terminal 30 of information. In this embodiment, the notification unit 106 notifies the mobile terminal 30 to input the input subjective data. The notification unit 106 may notify the mobile terminal 30 to input the input subjective data when the accuracy of the estimated subjective data output by the trained model 104 is equal to or lower than a predetermined threshold. The notification unit 106 may also notify the mobile terminal 30 to input the input subjective data when the accuracy of the estimated subjective data output by the trained model 104 is equal to or lower than a predetermined threshold for a predetermined period of time.
[0049] In the present embodiment, an example is shown in which the trained model 104 exists inside the information processing device 10, but the present disclosure is not limited to such an example. The trained model 104 may be generated and stored by a device other than the information processing device 10.
[0050] Next, a description will be given of the functional configuration of the mobile terminal 30. The continuous data acquisition unit 301 constantly acquires the odor sensing results sent from the odor sensor 20. The sensing results acquired by the continuous data acquisition unit 301 are assigned an acquisition date and time by the time assignment unit 305, and become sensor data.
[0051] The data transmission unit 302 acquires the sensor data from the continuous data acquisition unit 301 and transmits it to the information processing device 10 at all times.
[0052] The keyword input unit 303 allows the user to input subjective keywords about the odor of the space in which the odor sensor 20 is placed. The keyword input unit 303 presents a keyword input screen, which will be described later, and allows the user to select. The keyword input by the user through the keyword input unit 303 is given an acquisition date and time by the time assignment unit 305, and becomes input subjective data. Note that the number of subjective odor keywords that the keyword input unit 303 allows the user to input is not limited to one, and multiple keywords may be input. When inputting a keyword, the keyword input unit 303 may also allow the user to input information about the preferences of the user who inputs the keyword. Information about the user's preferences includes whether or not the user drinks alcohol and smokes.
[0053] The timing of inputting a keyword by the user through the keyword input unit 303 may be any timing, but in order for the information processing device 10 to effectively acquire input subjective data, the user may be made to input a keyword at the timing of acquiring the above-mentioned notification from the information processing device 10. Furthermore, when the user notices that the estimated subjective data generated by the information processing device 10 is clearly different from the odor in the space in which the odor sensor 20 is installed, the user may be able to input a keyword through the keyword input unit 303 by operating the mobile terminal 30 on the spot.
[0054] The data transmission unit 304 acquires the input subjective data from the keyword input unit 303, and transmits the input subjective data to the information processing device 10 at the timing of acquiring the input subjective data.
[0055] In addition, if the user is wearing a vital sensor such as a wearable device, the mobile terminal 30 may transmit the user's heart rate, blood pressure, body temperature, and other vital data measured at the same time as the input subjective data is transmitted to the information processing device 10. In addition, if the wearable device or the mobile terminal 30 is equipped with an environmental sensor that measures temperature, humidity, atmospheric pressure, altitude, and the like, the mobile terminal 30 may transmit environmental data measured at the same time as the input subjective data is transmitted to the information processing device 10. The information processing device 10 may use the vital data or environmental data transmitted from the mobile terminal 30 in generating the trained model 104 by machine learning.
[0056] The time assigning unit 305 assigns date and time information at the timing when the continuous data acquiring unit 301 acquires the sensing result and the timing when the keyword input unit 303 receives the input of a keyword from the user. The sensing data is data that is continuously and periodically measured. On the other hand, the input subjective data input by the user is data that is irregularly measured. In order to link the sensing data and the input subjective data, in this embodiment, the sensor data and the input subjective data are acquired in the same device, the mobile terminal 30, and the same time is used. In this embodiment, the time assigning unit 305 assigns date and time information, but the information assigned by the time assigning unit 305 is not limited to the example as long as it can link the sensing data and the input subjective data. For example, the time assigning unit 305 may assign a terminal-specific number and time information, or may assign unique ID information calculated by any calculation method, such as calculation of a hash value, from the terminal-specific number and time information.
[0057] The estimation utilization unit 306 executes processing based on the estimated subjective data of the odor in the space in which the odor sensor 20 is placed, which is estimated by the AI estimation unit 105 of the information processing device 10. The estimation utilization unit 306 also executes processing based on information related to the estimated subjective data notified by the notification unit 106 of the information processing device 10.
[0058] The estimation utilization unit 306 may, for example, execute a process of presenting the estimated subjective data of the odor in the space in which the odor sensor 20 is placed, estimated by the AI estimation unit 105, on the display unit 36. The estimation utilization unit 306 may also present, for example, on the display unit 36 a screen for prompting the user to input subjective data when notified by the notification unit 106 that the accuracy of the estimated subjective data in the space in which the odor sensor 20 is placed, estimated by the AI estimation unit 105, is equal to or lower than a predetermined threshold.
[0059] In this embodiment, the mobile terminal 30 to which the odor sensor 20 is connected and the mobile terminal 30 for inputting the subjective data are the same device, but the device to which the odor sensor 20 is connected and the device for inputting the subjective data may be different devices. However, when the device to which the odor sensor 20 is connected and the device for inputting the subjective data are different devices, it is desirable to ensure that the location where the odor sensor 20 is installed and the location where the subjective data is input are the same.
[0060] Next, the operation of the information processing device 10 will be described.
[0061] 5 is a flowchart showing the flow of the odor estimation process by the information processing device 10. The odor estimation process is performed by the CPU 11 reading out an odor estimation program from the ROM 12 or storage 14, deploying it in the RAM 13, and executing it.
[0062] In step S101, the CPU 11 estimates the subjective data of the odor from the sensor data sensed by the odor sensor 20 and transmitted from the mobile terminal 30, and generates estimated subjective data. When estimating the subjective data of the odor from the sensor data, the CPU 11 uses a trained model 104 that has been generated in advance by machine learning. Estimation using the trained model 104 refers to inputting sensor data to the trained model 104 and causing the trained model 104 to output estimated subjective data.
[0063] Following step S101, in step S102, the CPU 11 acquires input subjective data inputted via the mobile terminal 30 at the same evaluation date and time as the estimated subjective data estimated in step S101.
[0064] Following step S102, the CPU 11 determines in step S103 whether or not there is a difference between the estimated subjective data and the input subjective data.
[0065] If the result of the determination in step S103 is that there is no difference between the estimated subjective data and the input subjective data, that is, the estimated subjective data and the input subjective data match (step S103; Yes), CPU 11 ends the series of processes. Note that a match does not have to be a perfect match, and if multiple keywords are input as the input subjective data, a match may be determined if the input subjective data matches a predetermined amount or more.
[0066] On the other hand, if the determination result in step S103 indicates that there is a difference between the estimated subjective data and the input subjective data, that is, the estimated subjective data does not match the input subjective data (step S103; No), CPU 11 re-learns trained model 104 in step S104. Re-learning trained model 104 refers to adjusting parameters of trained model 104 so as to output subjective data that is the same as the input subjective data acquired in step S102, using the same sensor data as the sensor data used to output the estimated subjective data in step S101.
[0067] If there is a difference between the estimated subjective data and the input subjective data, the information processing device 10 re-learns the trained model 104, thereby improving the estimation accuracy of the estimated subjective data compared to a case where re-learning is not performed.
[0068] Next, there will be shown an example of a keyword input screen presented when the mobile terminal 30 inputs the input subjective data. Fig. 6A to Fig. 6D are diagrams showing an example of a keyword input screen presented when the mobile terminal 30 inputs the input subjective data at a predetermined timing.
[0069] The mobile terminal 30 may make the user input a predetermined number of keywords when selecting multiple keywords. The mobile terminal 30 may also make the user rank the keywords when selecting multiple keywords. The mobile terminal 30 may also select and present keywords from among keywords prepared in advance when presenting keywords. For example, if there are 100 keywords, the mobile terminal 30 may randomly select 20 keywords from the 100 keywords and present them.
[0070] When inputting subjective data, the mobile terminal 30 presents odor-related keywords prepared in advance and allows the user to select one. When the user selects "sweet" on the screen shown in Fig. 6A, the mobile terminal 30 makes the keyword "sweet" unselectable as shown in Fig. 6B. To make a keyword that has been selected unselectable, the mobile terminal 30 may mark it with a cross as shown in Fig. 6B, or may present the keyword in a different color from the other keywords to indicate that the keyword cannot be selected.
[0071] 6A to 6D, the mobile terminal 30 may display a screen for asking more detailed questions about the selected keyword. For example, when the keyword "sweet" is selected, sweet odors can be perceived in various ways, such as fruity odors, vanilla odors, heavy odors, and light odors. Therefore, the mobile terminal 30 may present a screen for asking more detailed questions about the type of sweet odor.
[0072] In this way, by presenting previously prepared keywords, the mobile terminal 30 can facilitate the input of subjective data regarding the subjective perception of odors and prevent variation in the keywords entered.
[0073] As described above, the information processing device 10 according to the embodiment of the present disclosure estimates the subjective data of the odor in the space in which the odor sensor 20 is placed, using the sensor data transmitted from the odor sensor 20 via the mobile terminal 30. Then, if there is a difference between the estimated subjective data and the actual input subjective data, the information processing device 10 according to the embodiment of the present disclosure re-trains the trained model 104 so as to estimate the actual input subjective data as the estimation result. If there is a difference between the estimated subjective data and the actual input subjective data, the information processing device 10 according to the embodiment of the present disclosure re-trains the trained model 104, thereby reducing the difference between the subjective data of the odor estimated by the AI and the subjective data of the odor estimated by humans.
[0074] Although the embodiments of the present disclosure have been described in detail above with reference to the accompanying drawings, the technical scope of the present disclosure is not limited to such examples. It is clear that a person having ordinary knowledge in the technical field of the present disclosure can conceive of various modified examples or examples of modifications within the scope of the technical ideas described in the claims, and it is understood that these modified examples or examples of modifications also belong to the technical scope of the present disclosure.
[0075] In addition, the effects described in the above embodiments are explanatory or exemplary, and are not limited to those described in the above embodiments. In other words, the technology according to the present disclosure may achieve other effects that are obvious to a person having ordinary skill in the technical field of the present disclosure from the description in the above embodiments, in addition to or in place of the effects described in the above embodiments.
[0076] In addition, the odor estimation process executed by the CPU by reading the software (program) in each of the above embodiments may be executed by various processors other than the CPU. In this case, examples of the processor include a PLD (Programmable Logic Device) such as an FPGA (Field-Programmable Gate Array) whose circuit configuration can be changed after manufacture, and a dedicated electric circuit such as an ASIC (Application Specific Integrated Circuit) which is a processor having a circuit configuration designed exclusively for executing a specific process. In addition, the odor estimation process may be executed by one of these various processors, or may be executed by a combination of two or more processors of the same or different types (for example, a plurality of FPGAs, a combination of a CPU and an FPGA, etc.). In addition, the hardware structure of these various processors is, more specifically, an electric circuit that combines circuit elements such as semiconductor elements.
[0077] In addition, in each of the above embodiments, the odor estimation processing program is pre-stored (installed) in a ROM or storage, but the present invention is not limited to this. The program may be provided in a form recorded on a non-transitory recording medium such as a CD-ROM (Compact Disk Read Only Memory), a DVD-ROM (Digital Versatile Disk Read Only Memory), or a USB (Universal Serial Bus) memory. The program may also be downloaded from an external device via a network. [Explanation of symbols]
[0078] 10. Information processing device 20 Odor Sensor 30 Mobile Devices 40 Display device
Claims
1. an estimation unit that uses a trained model for estimating subjective data of an odor from sensor data obtained by continuously sensing an odor, the sensor data being continuously transmitted from the odor sensor through a terminal, to obtain estimated subjective data of the odor; a learning unit that, if there is a difference between the estimated subjective data at the evaluation date and time in input subjective data input to the terminal by a user who smelled the odor and information on the evaluation date and time when the odor was evaluated, and the input subjective data, re-learns the trained model so as to output the input subjective data by inputting the sensor data corresponding to the odor; An odor estimation device comprising:
2. The odor estimation device according to claim 1 , further comprising a notification unit that notifies the user to input the input subjective data into the terminal.
3. 3. The odor estimation device according to claim 2, wherein the notification unit notifies the terminal to input the input subjective data when the accuracy of the estimated subjective data output by the trained model is equal to or lower than a predetermined threshold.
4. 4. The odor estimation device according to claim 3, wherein the notification unit notifies the terminal to input the input subjective data when a time period during which the accuracy of the estimated subjective data output by the trained model is equal to or lower than a predetermined threshold continues for a predetermined period of time.
5. The odor estimation device according to claim 1 , wherein the sensor data includes information on the time at which the odor sensor performed sensing.
6. an odor sensor that continuously senses odors and obtains sensor data; a terminal connected to the odor sensor to acquire the sensor data; an odor estimation device that acquires sensor data from the terminal via wireless communication; Equipped with The odor estimation device comprises: an estimation unit that uses a trained model for estimating subjective data of an odor from sensor data obtained by continuously sensing an odor, the sensor data being continuously transmitted from an odor sensor through a terminal, to obtain estimated subjective data of the odor; a learning unit that, if there is a difference between the estimated subjective data at the evaluation date and time in input subjective data input to the terminal by a user who smelled the odor and information on the evaluation date and time when the odor was evaluated, and the input subjective data, re-learns the trained model so as to output the input subjective data by inputting the sensor data corresponding to the odor; An odor estimation system comprising:
7. The processor: Obtaining estimated subjective data of the odor from sensor data obtained by continuously sensing the odor, which is continuously transmitted from the odor sensor through the terminal, using a trained model that estimates subjective data of the odor; If there is a difference between the estimated subjective data at the evaluation date and time in input subjective data input to the terminal by a user who smelled the odor, the input subjective data including information on the subjectivity of the odor and the evaluation date and time when the odor was evaluated, the trained model is retrained so as to output the input subjective data by inputting the sensor data corresponding to the odor. An odor estimation method for performing processing.
8. On the computer, Obtaining estimated subjective data of the odor from sensor data obtained by continuously sensing the odor, which is continuously transmitted from the odor sensor through the terminal, using a trained model that estimates subjective data of the odor; If there is a difference between the estimated subjective data at the evaluation date and time in input subjective data input to the terminal by a user who smelled the odor, the input subjective data including information on the subjectivity of the odor and the evaluation date and time when the odor was evaluated, the trained model is retrained so as to output the input subjective data by inputting the sensor data corresponding to the odor. An odor estimation program that executes the process.
Citation Information
Patent Citations
Information management system, information management method, and program
WO2021171987A1
Information processing device and program
WO2022176923A1
