JP2023077288A - Data collection device, data collection method, and terminal device - Google Patents

Abstract

To prevent data from excessively stored in a data collection device while suppressing deterioration of learning accuracy of a machine learning model.SOLUTION: A data collection device collects data of a parameter used for learning of a machine learning model which outputs an output parameter regarding occurrence of a specific event for an object from a terminal apparatus 10. The data collection device includes: a determination unit 231 which determines whether a specific event has ever occurred in the object which holds the terminal apparatus; and a data control unit 232 which controls a data volume of each parameter which are acquired by the terminal apparatus and stored in a storage device per unit time. The data control unit controls the data volume so that a data volume of each of parameters which are acquired by terminal apparatuses held by objects in which a specific event has ever occurred and stored in the storage device is larger than a data volume of each of parameters which are acquired by terminal apparatuses held by objects in which a specific event has never occurred and stored in the storage device.SELECTED DRAWING: Figure 6

Description

The present disclosure relates to a data collection device, a data collection method, and a terminal device.

In smart cities, it is proposed to collect data from multiple entities within the community. In particular, Patent Document 1 proposes to collect data obtained by correcting the obtained data, since there is uncertainty in the data obtained from the information systems of different business entities, in order to eliminate such uncertainty. ing.

JP 2013-069084 A

By the way, in general, in order to improve the calculation accuracy of a machine learning model, a large amount of data for performing machine learning is required. On the other hand, in a data collection device that collects data for use in learning a machine learning model from terminal devices (mobile terminals, etc.) located in a smart city, data is transmitted from many terminal devices. Therefore, if all the transmitted data is stored in the storage device of the data collection device, the amount of data stored in the storage device becomes too large.

In view of the above problems, an object of the present disclosure is to suppress excessive storage of data in a data collection device while suppressing deterioration in learning accuracy of a machine learning model.

The gist of the present disclosure is as follows.

(1) A data collection device that collects, from a terminal device, parameter data used for learning a machine learning model that outputs an output parameter related to the occurrence of a specific event about an object,
a storage device for storing data;
a determination unit that determines whether or not the specific event has occurred in the object holding the terminal device;
a data control unit that controls the amount of data of each parameter acquired by the terminal device and stored in the storage device per unit time;
The data control unit controls the amount of data of each parameter acquired by the terminal device held by the object in which the specific event has occurred and stored in the storage device so that the amount of data of each parameter obtained by the specific event has occurred. A data collection device that controls the amount of data so that the amount of data is greater than the amount of data of each parameter that is acquired by the terminal device held by the object and stored in the storage device.
(2) The data control unit adjusts the data to be stored in the storage device among the data of each parameter transmitted from the terminal device to the data collection device, thereby adjusting the data acquired by the terminal device and stored in the storage device. The data collection device according to (1) above, which controls the amount of data of each parameter stored in the .
(3) The data control unit adjusts the data transmitted to the data collection device among the data of each parameter detected by the terminal device, so that the data acquired by the terminal device and stored in the storage device is The data collection device according to (1) or (2) above, which controls the amount of data for each parameter.
(4) The data control unit adjusts the data acquired from the detector among the data of each parameter that can be detected by the detector of the terminal device, so that the data acquired by the terminal device and stored in the storage device The data collection device according to any one of (1) to (3) above, which controls the amount of data for each parameter to be collected.
(5) said output parameter is the probability of occurrence of said particular event;
When the probability calculated in the past is relatively high, the data control unit acquires by the terminal device and stores in the storage device more than when the probability calculated in the past is relatively low. The data collection device according to any one of (1) to (4) above, wherein the data amount is controlled so that the data amount of each parameter to be collected is increased.
(6) A data collection method for collecting, from a terminal device, parameter data used for learning a machine learning model that outputs output parameters relating to the occurrence of a specific event about an object,
Determining whether the specific event has occurred in an object holding the terminal device;
controlling the amount of data of each parameter acquired by the terminal device and stored in the storage device of the data collection device per unit time;
The amount of data of each parameter acquired by the terminal device held by the object in which the specific event has occurred and stored in the storage device is obtained by the object in which the specific event has never occurred A data collection method, wherein the amount of data of each parameter obtained by the held terminal device and stored in the storage device is larger.
(7) A terminal device that transmits parameter data used for learning a machine learning model that outputs output parameters related to the occurrence of a specific event about an object to a data collection device,
a determination unit that determines whether or not the specific event has occurred in the object holding the terminal device;
a data control unit that controls the amount of data of each parameter transmitted from the terminal device to the data collection device per unit time;
The data control unit controls the amount of data of each parameter to be acquired and transmitted to the data collection device when the terminal device is held by an object on which the specific event has occurred. A terminal device that controls the amount of data to be greater than the amount of data of each parameter acquired and transmitted to the data collection device when held by an object that has never experienced the specific event. .

According to the present disclosure, excessive storage of data in a data collection device is suppressed while suppressing a decrease in learning accuracy of a machine learning model.

FIG. 1 is a schematic configuration diagram of a machine learning system. FIG. 2 is a diagram schematically showing the hardware configuration of the terminal device. FIG. 3 is a functional block diagram of the processor of the terminal device. FIG. 4 is a diagram schematically showing the hardware configuration of the server. FIG. 5 is a functional block diagram of the processor of the server. FIG. 6 is a sequence diagram showing the flow of data collection processing for use in machine learning model learning. FIG. 7 is a sequence diagram, similar to FIG. 6, showing the flow of data collection processing for use in learning a machine learning model. FIG. 8 is a sequence diagram showing the flow of machine learning model learning processing in the server. FIG. 9 is a functional block diagram of the processor of the terminal device; FIG. 10 is a sequence diagram showing the flow of data collection processing for use in machine learning model learning.

Hereinafter, embodiments will be described in detail with reference to the drawings. In the following description, the same reference numerals are given to the same constituent elements.

・First embodiment <Configuration of machine learning system>
The configuration of a machine learning system 1 according to the first embodiment will be described with reference to FIGS. 1 to 5. FIG. FIG. 1 is a schematic configuration diagram of a machine learning system 1. As shown in FIG. In the machine learning system 1, machine learning models are learned in the server. The machine learning system 1 also functions as a data collection system that collects data necessary for learning a machine learning model.

As shown in FIG. 1 , the machine learning system 1 has a plurality of terminal devices 10 and a server 20 that can communicate with the terminal devices 10 . Each of the plurality of terminal devices 10 and the server 20 are connected via a communication network 4 configured by an optical communication line or the like, and a wireless base station 5 connected to the communication network 4 via a gateway (not shown). , are configured to communicate with each other. For communication between the terminal device 10 and the radio base station 5, various wide-area radio communication with a long communication distance can be used. Standards-compliant communication is used.

In particular, in this embodiment, the server 20 communicates with terminal devices 10 located within a predetermined target area. The target area is a range surrounded by a predetermined boundary. A smart city is defined as a sustainable city or region that solves the problems it faces and continues to create new value. The server 20 may be able to communicate with the terminal device 10 located outside the target area.

The terminal device 10 acquires data necessary for processing using a machine learning model and learning of the machine learning model, which will be described later. In particular, in the present embodiment, the terminal device 10 is a device that is held by an individual and obtains data of the individual holding the terminal device 10 . Therefore, in the present embodiment, the terminal device 10 functions as a mobile data acquisition device that acquires personal data within a predetermined target area or its surrounding area. Therefore, in this embodiment, the terminal device 10 moves along with the movement of the person holding the terminal device 10 .

Specifically, in the present embodiment, the terminal device 10 is, for example, a watch-type terminal (smart watch), a wristband-type terminal, a clip-type terminal, a wearable terminal such as a glasses-type terminal (smart glasses), and a mobile terminal. including. Such a terminal device 10 acquires, for example, position information of each person in the target area and personal data regarding the state of the user wearing the terminal device 10 . Personal data includes biometric data such as vital signs (heart rate, body temperature, blood pressure and respiratory rate), blood oxygen concentration, electrocardiogram, blood sugar level, step count, calorie consumption, fatigue level, and sleep state.

In addition, in this embodiment, the terminal device 10 particularly includes a watch-type terminal and a mobile terminal that communicates with the watch-type terminal by short-range wireless communication. As near-field wireless communication, for example, communication conforming to any communication standard established by IEEE, ISO, IEC, etc. (for example, Bluetooth (registered trademark), ZigBee (registered trademark)) is used.

FIG. 2 is a diagram schematically showing the hardware configuration of the terminal device 10. As shown in FIG. As shown in FIG. 2 , terminal device 10 has communication module 11 , sensor 12 , input device 13 , output device 14 , memory 15 and processor 16 . Communication module 11, sensor 12, input device 13, output device 14 and memory 15 are connected to processor 16 via signal lines.

The communication module 11 is an example of a communication device that communicates with other devices. The communication module 11 is, for example, equipment for communicating with the server 20 . In particular, the communication module 11 is a device that communicates with the wireless base station 5 by the wide area wireless communication described above.

Sensor 12 is an example of a detector that detects various parameters related to the individual holding terminal device 10 . Further, the sensor 12 detects various parameters such as parameters related to the situation of the terminal device 10 and the situation around the terminal device 10 . In particular, sensor 12 has multiple individual sensors that detect different parameters. Values of various parameters detected by the sensor 12 are transmitted to the processor 16 or memory 15 via signal lines.

Specifically, the sensors 12 include sensors that detect parameters related to the user holding the terminal device 10 . For example, if the terminal device 10 is a watch-type terminal (smartwatch), the sensor 12 includes a sensor that detects personal data (including biometric data) of the user wearing the terminal device 10 . Additionally, the sensors 12 include sensors that detect the status of the terminal device 10 , such as GNSS receivers that detect the current location of the terminal device 10 . The sensor 12 may also include a sensor that detects environmental data around the terminal device 10 .

The input device 13 is a device for the user of the terminal device 10 to make an input. Specifically, the input device 13 includes a touch panel, a microphone, buttons, dials, and the like. Information input via the input device 13 is transmitted to the processor 16 or the memory 15 via signal lines.

The output device 14 is a device for the terminal device 10 to output. Specifically, the output device 14 includes a display, speakers, and the like. The output device 14 outputs based on commands sent from the processor 16 via signal lines. For example, the display causes an image to be displayed on the screen based on instructions from the processor 16 and the speaker outputs sound based on instructions from the processor 16 .

The memory 15 has, for example, a volatile semiconductor memory (eg, RAM), a non-volatile semiconductor memory (eg, ROM), and the like. The memory 15 stores computer programs for executing various processes in the processor 16, parameter data detected by the sensor 12, data related to the machine learning model (e.g. configuration of the machine learning model and learning parameters (e.g. weights, biases, etc.). ) data), various data used when various processes are executed by the processor 16, and the like are stored.

The processor 16 has one or more CPUs (Central Processing Units) and their peripheral circuits. Processor 16 may further comprise arithmetic circuitry such as a logic arithmetic unit or a math arithmetic unit. The processor 16 executes various processes based on computer programs stored in the memory 15 . Specific processing executed by the processor 16 of the terminal device 10 will be described later.

FIG. 3 is a functional block diagram of the processor 16 of the terminal device 10. As shown in FIG. As shown in FIG. 3 , the processor 16 of the terminal device 10 has a data acquisition section 161 , a model execution section 162 , a notification section 163 , a data transmission section 164 and a model update section 165 . The data acquisition unit 161 acquires various data from the sensor 12 or the like. The model execution unit 162 uses the machine learning model to calculate the values of the output parameters. The notification unit 163 notifies the user based on the calculated value of the output parameter. The data transmission unit 164 transmits data from the terminal device 10 to the server 20 . The model update unit 165 updates values of learning parameters of the machine learning model used in the model execution unit 162 . These functional blocks of the processor 16 of the terminal device 10 are, for example, functional modules implemented by computer programs running on the processor 16 . Alternatively, these functional blocks of the processor 16 may be dedicated arithmetic circuits provided in the processor 16 . Details of each of these functional blocks will be described later.

The server 20 is connected to multiple terminal devices 10 via the communication network 4 . In this embodiment, the server 20 executes processing using a machine learning model and learns the machine learning model. The server 20 also functions as a data collection device that collects data necessary for execution of processing using a machine learning model and learning of the machine learning model.

FIG. 4 is a diagram schematically showing the hardware configuration of server 20. As shown in FIG. The server 20 includes a communication module 21, a storage device 22, and a processor 23, as shown in FIG. The server 20 may also have input devices such as a keyboard and mouse, and output devices such as a display and speakers.

The communication module 21 is an example of a communication device that communicates with equipment outside the server 20 . Communication module 21 comprises an interface circuit for connecting server 20 to communication network 4 . The communication module 21 is configured to communicate with each of the plurality of terminal devices 10 via the communication network 4 and the radio base station 5 .

The storage device 22 is an example of a storage device that stores data. The storage device 22 has, for example, a hard disk drive (HDD), a solid state drive (SSD), or an optical recording medium. Also, the storage device 22 may have a volatile semiconductor memory (eg, RAM), a non-volatile semiconductor memory (eg, ROM), or the like. The storage device 22 stores a computer program for executing various processes by the processor 23 and various data used when the processor 23 executes various processes. In particular, the storage device 22 stores data received from the terminal device 10, data related to the machine learning model, and data used for learning the machine learning model.

The processor 23 has one or more CPUs and their peripheral circuits. The processor 23 may further comprise a GPU or an arithmetic circuit such as a logical arithmetic unit or a numerical arithmetic unit. The processor 23 executes various processes based on computer programs stored in the storage device 22 .

FIG. 5 is a functional block diagram of the processor 23 of the server 20. As shown in FIG. As shown in FIG. 5 , the processor 23 has a determination section 231 , a data control section 232 , a data transmission section 233 , a data set creation section 234 and a learning section 235 . The determination unit 231 determines whether or not the user holding the terminal device 10 has had a specific disease. The data control unit 232 controls the amount of data of each parameter acquired by the terminal device 10 and stored in the storage device 22 . The data transmission unit 233 transmits data from the server 20 to the terminal device 10 . The dataset creating unit 234 creates a learning dataset. The learning unit 235 uses the created data set to learn a machine learning model. These functional blocks possessed by the processor 23 of the server 20 are, for example, functional modules implemented by computer programs running on the processor 23 . Alternatively, these functional blocks of the processor 23 may be dedicated arithmetic circuits provided in the processor 23 . Details of each of these functional blocks will be described later.

<Machine learning model>
In this embodiment, a machine-learned model is used when performing predetermined processing in the terminal device 10 . In this embodiment, the machine learning model is a model that outputs output parameters related to the occurrence of a specific event regarding the object based on data obtained from the sensor 12 of the terminal device 10 and the like.

Specifically, in the present embodiment, personal data of the user holding the terminal device 10 is input to the machine learning model. Further, environmental data around the terminal device 10 may be input to the machine learning model. The personal data and environmental data of the user holding the terminal device 10 are acquired from the sensor 12 of the terminal device 10 . Alternatively, environmental data may be obtained from the server 20 via the communication module 11 instead of from the sensor 12 . Then, when the data of such input parameters are input, the machine learning model outputs the probability that the person holding the terminal device 10 will develop a specific disease (arrhythmia, heat stroke, etc.).

Various machine learning algorithms can be used for the machine learning model. In this embodiment, the machine learning model is a model learned by supervised learning, such as neural network (NN), support vector machine (SVM), decision tree (DT). In particular, in this embodiment, the machine learning model is preferably a recurrent neural network (RNN) model that inputs the user's personal data and environmental data in chronological order as input parameters.

Note that models having various input parameters and output parameters can be used as machine learning models. Input parameters include various parameters that can be detected by the sensor 12 of the terminal device 10 . Specifically, the input parameters include, for example, vital signs (heart rate, body temperature, blood pressure and respiratory rate), blood oxygen concentration, electrocardiogram, blood sugar level, number of steps, calorie consumption, fatigue level, sleep state, time, image , videos, etc. The input parameters may also include parameters transmitted from the server 20 via the communication network 4 (for example, temperature, humidity, weather, wind speed, etc. around the terminal device 10). In addition, the output parameters include various parameters relating to the occurrence of specific events for the user holding terminal device 10 .

In this embodiment, the machine learning of the machine learning model as described above is performed by the server 20 instead of the terminal device 10 . A machine learning model is trained using a training dataset. The learning data set includes data used as input parameters and output parameter values (correct values or correct labels) corresponding to this data. In particular, in the present embodiment, the learning data set includes time-series data acquired by the terminal device 10 for a certain subject, and data as to whether or not a specific disease has occurred in the subject. . For example, when the probability of occurrence of arrhythmia is the output parameter, the probability of occurrence of arrhythmia, which is the output parameter, is set to 1 in the learning data set created for a person who has had an arrhythmia. On the other hand, the probability of occurrence of arrhythmia, which is an output parameter, is set to 0 in the learning data set created for persons who did not experience arrhythmia. Also, the learning data set may be generated by performing preprocessing (loss processing, normalization, standardization, etc.) on the output values of the sensor 12 .

In learning the machine learning model, for example, learning parameters (NN weight w, bias b, etc.) in the machine learning model are repeatedly updated by any known method (eg, error backpropagation method). The learning parameters are repeatedly updated, for example, so that the difference between the output values of the machine learning model and the correct values of the output parameters included in the learning data set is reduced. As a result, the machine learning model is learned and a learned machine learning model is generated.

<Processing using a machine learning model>
Next, processing using a machine learning model in the terminal device 10 will be described with reference to FIG. In this embodiment, the terminal device 10 estimates the probability that the user holding the terminal device 10 will develop a specific disease based on the values of various parameters detected by the sensor 12 of the terminal device 10 . In addition, the terminal device 10 notifies the user that there is a risk of contracting a specific disease when the probability of developing the specific disease is greater than or equal to a predetermined reference value. The processor 16 of the terminal device 10 uses a data acquisition unit 161, a model execution unit 162, and a notification unit 163 to estimate the probability of occurrence of a specific disease.

The data acquisition unit 161 acquires data including data regarding input parameters of the machine learning model. Specifically, the data acquisition unit 161 acquires the values of the input parameters detected by the sensor 12 of the terminal device 10 from the memory 15 . In this embodiment, the data acquisition unit 161 acquires the user's personal data, environment data, and the like from the sensor 12 . The data acquisition unit 161 may also acquire input parameter values from an external device such as the server 20 via the communication module 11 . In this embodiment, when the server 20 receives the current position detected by the sensor 12 of each terminal device 10 from each terminal device 10, the server 20 transmits the current temperature and current humidity around that position to the terminal device 10. do. Therefore, the data acquisition unit 161 acquires the temperature and humidity around the terminal device 10 from the server 20 . The data acquired by the data acquisition unit 161 is stored in the memory 15 .

When the data acquisition unit 161 acquires the current values of the input parameters of the machine learning model, the model execution unit 162 inputs the acquired values of the input parameters to the machine learning model and calculates the values of the output parameters. In this embodiment, when the user's personal data and environmental data acquired by the data acquisition unit 161 are input to the machine learning model, the model execution unit 162 outputs the probability that the user will develop a specific disease. .

Here, the program for executing the machine learning model and the values of the learning parameters used in the machine learning model are stored in the memory 15 . Therefore, the model execution unit 162 uses the programs stored in the memory 15 and the values of the learning parameters to calculate the values of the output parameters.

The notification unit 163 notifies the user via the output device 14 based on the output parameter values calculated by the model execution unit 162 . Specifically, in the present embodiment, when the probability that the user will develop a specific disease calculated by the model execution unit 162 is equal to or greater than a predetermined reference value, the notification unit 163 notifies the user through the output device 14. notification. In this case, the notification unit 163 may, for example, display a warning regarding the specific disease on the display, or generate a warning sound regarding the specific disease from the speaker.

<Collection of learning data>
Next, with reference to FIGS. 3 and 5 to 7, data collection processing for use in machine learning model learning will be described. In this embodiment, learning of the machine learning model is performed in the server 20 based on data acquired in the terminal device 10 and the like. Therefore, in the present embodiment, data acquired by the terminal device 10 is transmitted to the server 20 and stored in the storage device 22 of the server 20 when the machine learning model is learned.

The processor 16 of the terminal device 10 uses the data transmission unit 164 to perform data collection processing. Also, the processor 23 of the server 20 uses the determination unit 231, the data control unit 232, and the data transmission unit 233 in performing the data collection process.

By the way, in learning a machine learning model for estimating the probability that a user will develop a specific disease based on the data acquired from the terminal device 10, the data acquired by the terminal device 10 is transmitted to the server 20. . However, when all the data acquired by all the terminal devices 10 are transmitted to the server 20 and stored in the storage device 22 of the server 20, the storage device 22 stores a large amount of data. Therefore, a storage device 22 with a very large storage capacity is required.

On the other hand, the probability that a user holding each terminal device 10 will develop a specific disease is low. Therefore, if the data acquired by all the terminal devices 10 are used for learning the machine learning model, the data obtained when the user does not have a specific disease becomes excessive. Even if there is an excess of data when the user does not have a specific disease, the estimation accuracy of the machine learning model cannot necessarily be improved.

On the other hand, users who have had a particular disease in the past are more likely to develop that particular disease than other users. Therefore, in the present embodiment, the amount of data of each parameter acquired by the terminal device 10 held by a user who has had a specific disease in the past and stored in the storage device 22 of the server 20 is is larger than the amount of data of each parameter acquired by the terminal device 10 held by the user who has never had the disease and stored in the storage device 22 .

FIG. 6 is a sequence diagram showing the flow of data collection processing for use in machine learning model learning. The learning process shown in FIG. 6 is executed by the processor 16 of the terminal device 10 and the processor 23 of the server 20 .

In the data collection process in this embodiment, as shown in FIG. 6, first, the data transmission unit 164 of the processor 16 of each terminal device 10 transmits identification information to the server 20 (step S11). The identification information is information for identifying the user who owns the terminal device 10, and is, for example, a user ID or an e-mail address. The identification information is input in advance by the user through the input device 13 and stored in the memory 15 . Therefore, data transmission unit 164 transmits the identification information stored in memory 15 to server 20 . The data transmission unit 164 transmits the identification information, for example, when new identification information is input, or when a certain period of time has passed since the previous transmission of the identification information.

When the server 20 receives the identification information from the terminal device 10, the determination unit 231 of the server 20 determines whether or not the user identified by the identification information has had a specific disease in the past (step S12). The storage device 22 of the server 20 stores a list of identification information of users who have had specific diseases in the past. The storage device 22 may also store other related information such as the number of times a particular disease has occurred in each user. Therefore, the determination unit 231 collates the identification information received from the terminal device 10 with the list to determine whether or not the user holding the terminal device 10 has had a specific disease in the past. In addition, the determination unit 231 may acquire other related information such as the number of occurrences of a user who has had a specific disease in the past.

The identification information of a person with a specific disease is transmitted to the server 20 through various routes. For example, the user's identification information and disease information are input to a terminal device (not shown) provided at a medical institution that has examined a user suffering from a specific disease, and the input information is transmitted to the server 20. . Alternatively, a user suffering from a specific disease inputs the disease information to the terminal device 10 , and the input information is transmitted to the server 20 . The server 20 creates a list of users who have had a specific disease in the past, based on the information thus received from the terminal device or terminal device 10 of the medical institution.

When it is determined whether or not the user has developed a specific disease, the data control unit 232 of the server 20 transmits data from the terminal device 10 to the server 20 per unit time based on the determination result. A target data amount of the parameter is set (step S13). The data control unit 232 may set the target data amount for each parameter, or may collectively set the target data amount for each of all the parameters. The target data amount is set with the maximum value of the data amount that can be acquired by the terminal device 10 from the sensor 12 or the like as the upper limit.

In this embodiment, the data control unit 232 determines that the amount of data to be transmitted from the terminal device 10 held by a user who has had a specific disease in the past to the server 20 is the The target data amount is set so as to be larger than the data amount to be transmitted from the terminal device 10 held by a non-user to the server 20 .

When the target data amount is set, the data transmission unit 233 of the server 20 transmits the set target data amount to the terminal device 10 (step S14). The target data amount received by the terminal device 10 is stored in the memory 15 of the terminal device 10 .

On the other hand, the sensor 12 of the terminal device 10 detects personal data and environmental data at arbitrary time intervals. The data acquisition unit 161 of the processor 16 of the terminal device 10 acquires the personal data and environmental data thus detected by the sensor 12 from the sensor 12 (step S15). Note that the data acquisition unit 161 may acquire from the memory 15 the data detected by the sensor 12 and then stored in the memory 15 .

The data transmission unit 164 of the terminal device 10 transmits the acquired data to the server 20 (step S16). In particular, in this embodiment, the data transmission unit 164 transmits data according to the target data amount transmitted from the server 20 in step S14. In particular, in this embodiment, the data transmission unit 164 controls the amount of data transmitted to the server 20 by adjusting the data transmitted to the server 20 among the data of each parameter detected by the terminal device 10. .

Therefore, the data transmission unit 164 transmits all of the acquired data to the server 20 when the target data amount is large. On the other hand, the data transmission unit 164 transmits part of the acquired data to the server 20 when the target data amount is small. For example, when the target data volume is half of the maximum data volume that can be transmitted, the data transmission unit 164 transmits the data acquired in chronological order for each parameter to the server 20 every other time. The data transmitted to the server 20 in this manner is stored in the storage device 22 of the server 20 (step S17).

In this embodiment, the data transmission unit 164 controls the data amount of each parameter to be transmitted to the server 20 by adjusting the data transmitted to the server 20 among the data of each parameter detected by the terminal device 10. , thus controlling the amount of data stored in the storage device 22 . Therefore, in this embodiment, the data control unit 232 adjusts the data amount of each parameter acquired by the terminal device 10 and stored in the storage device 22 per unit time by adjusting the data transmitted to the server 20. can be said to be in control.

Note that, in this embodiment, the data amount of each parameter stored in the storage device 22 may be controlled by other methods. Specifically, for example, the data amount of each parameter stored in the storage device 22 may be controlled by adjusting the data acquired from the sensor 12 among the data of each parameter detectable by the sensor 12 . In this case, for example, the data acquisition unit 161 acquires data detected by the sensor 12 every 10 ms from the sensor 12 that can detect parameter values every 10 ms when the target data amount is large. On the other hand, the data acquisition unit 161 acquires the data detected by the sensor 12 every 20 ms from the sensor 12 that can detect the value of the parameter every 10 ms when the target data amount is small. In this case, the data control unit 232 adjusts the data acquired from the sensor 12 among the data of each parameter that can be detected by the sensor 12 of the terminal device 10, so that the data acquired by the terminal device 10 and stored in the storage device 22 per unit time. It can be said that the amount of data stored for each parameter is controlled per unit.

Alternatively, the data amount of each parameter stored in the storage device 22 may be controlled by adjusting the data to be stored in the storage device 22 among the data of each parameter transmitted from the terminal device 10 to the server 20 . FIG. 7 is a sequence diagram, similar to FIG. 6, showing the flow of data collection processing for use in learning a machine learning model. In the example shown in FIG. 7, the data transmission unit 233 does not transmit the target data amount to the terminal device 10 even if the target data amount is set. Therefore, all the data of each parameter detectable by the sensor 12 are acquired by the data acquisition unit 161 and transmitted to the server 20 by the data transmission unit 164 (steps S15 and S16). The data transmitted to the server 20 is then stored in the storage device 22 of the server 20 (step S18). At this time, based on the target data amount calculated in step S13, the data to be stored in the storage device 22 among the data of each parameter transmitted from the terminal device 10 to the server 20 is adjusted, thereby storing in the storage device 22. Controls the amount of data in the parameters that are used. Therefore, in this case, the data control unit 232 adjusts the data of each parameter transmitted from the terminal device 10 to the server 20 to be stored in the storage device 22 per unit time. It can be said that the amount of data for each parameter is controlled.

<Learning machine learning model>
Next, the learning process of the machine learning model used in the server 20 will be described with reference to FIG. FIG. 8 is a sequence diagram showing the flow of machine learning model learning processing in the server 20 . For the learning process, the data set creation unit 234 and the learning unit 235 of the server 20 and the model updating unit 165 of the terminal device 10 are used.

When the data (personal data, environmental data, etc.) transmitted from each terminal device 10 is stored to some extent in the storage device 22, the dataset creation unit 234 creates a learning dataset (step S21). The learning data set includes measured values of input parameters of the machine learning model and correct data (correct values or correct labels) of output parameters. For example, in the present embodiment, the learning data set includes data acquired by the terminal device 10 held by each user and information (correct data) regarding the occurrence of diseases of the user.

Data acquired by the terminal device 10 held by each user is stored in the storage device 22 of the server 20 . Therefore, the data set creation unit 234 uses the data stored in the storage device 22 in this manner to create the learning data set.

Further, in this embodiment, when each user suffers from a specific disease, the information is input to a terminal device provided in a medical institution. Disease information input to the terminal device of the medical institution is transmitted to the server 20 via the communication network 4 . The data set creation unit 234 uses such disease information as correct data when creating a learning data set for the machine learning model. Note that, if the user has input disease information into the terminal device 10, the data set creation unit 234 may use the disease information as correct data.

When a certain number of learning data sets are created by the data set creation unit 234 in step S31, the learning unit 235 uses the created data sets to learn a machine learning model (step S22). Specifically, as described above, the learning unit 235 updates the learning parameters used in the machine learning model using a known error backpropagation method or the like.

When learning of the machine learning model is completed, the learning unit 235 transmits learning parameters of the machine learning model that has been learned to each terminal device 10 (step S23). Specifically, the learning unit 235 transmits the value of the learning parameter of the machine learning model that has been learned to each terminal device 10 .

When the model updating unit 165 of the terminal device 10 receives the learned machine learning model from the server 20, it updates the value of the learning parameter of the machine learning model for estimating the probability of occurrence of a specific disease (step S33). After the values of the learning parameters of the machine learning model are updated, the machine learning model using the updated learning parameters estimates the probability of developing a specific disease.

<Effect/Modification>
In the above embodiment, the data amount of each parameter acquired by the terminal device 10 held by the user who has had a specific disease in the past and stored in the storage device 22 is The data amount of each parameter is set larger than the data amount of each parameter that is acquired by the terminal device 10 held by the user who does not use the data and stored in the storage device 22 . Since the amount of data of each parameter acquired by the terminal device 10 held by the user who has had a specific disease in the past and stored in the storage device 22 is large, the data stored in this way can be used Even if the machine learning model is learned, the learning accuracy of the machine learning model can be relatively high. On the other hand, since the data amount of each parameter acquired by the terminal device 10 held by the user who has never had a specific disease in the past and stored in the storage device 22 is reduced, the data amount stored in the storage device 22 is reduced. The amount of data to be processed can be reduced. Therefore, according to the above-described embodiment, it is possible to suppress excessive storage of data in the data collection device while suppressing deterioration in the learning accuracy of the machine learning model.

It should be noted that in the above embodiment, a model that outputs output parameters relating to the occurrence of a specific disease for a user is used as the machine learning model. However, the machine learning model may be a model relating to objects unrelated to humans, or a model that outputs output parameters relating to the occurrence of events unrelated to disease. For example, the machine learning model may be a model that outputs output parameters related to the occurrence of vehicle failures. In this case, the vehicle's electronic control unit functions as a terminal device held in the vehicle.

Further, in the above-described embodiment, the terminal device 10 performs the process using the machine learning model (the process of estimating the probability of occurrence of a specific disease). However, such processing may be performed at the server 20 . In this case, all data relating to the input parameters of the machine learning model are transmitted from the terminal device 10 to the server 20 , and the calculated probability of occurrence of the specific disease is transmitted from the server 20 to the terminal device 10 .

Furthermore, in the above embodiment, the amount of data of each parameter acquired by the terminal device 10 held by a user who has never had a specific disease in the past and stored in the storage device 22 is the same as that of the user who has a specific disease. is assumed to be constant regardless of the probability of occurrence of However, the data amount of each parameter stored in the storage device 22 may be changed according to the probability that the user will develop a specific disease calculated using a machine learning model in the past. In this case, specifically, the data control unit 232, when the probability of occurrence of a specific disease calculated in the past is relatively high, is higher than when the probability calculated in the past is relatively low. , the data amount is controlled so that the data amount of each parameter acquired by the terminal device 10 and stored in the storage device 22 is increased.

- Second embodiment Next, a machine learning system 1 according to a second embodiment will be described with reference to Figs. 9 and 10 . Since the configuration and control in the machine learning system according to the second embodiment are the same as the configuration and control in the machine learning system according to the first embodiment, the points different from the machine learning system according to the first embodiment will be described below. Mainly explained.

In the first embodiment described above, the determination unit 231 of the server 20 determines whether or not the user has had a specific disease, and the data control unit 232 of the server 20 sets the target data amount. In contrast, in the present embodiment, the determination unit 166 of the terminal device 10 determines whether or not the user has a specific disease, and the data control unit 167 of the terminal device 10 sets the target data amount. do.

FIG. 9 is a functional block diagram of the processor 16 of the terminal device 10 according to the second embodiment. As shown in FIG. 9, the processor 16 of the terminal device 10 has a determination section 166 and a data control section 167 in addition to each section in the first embodiment.

FIG. 10 is a sequence diagram showing the flow of data collection processing for use in machine learning model learning. The learning process shown in FIG. 10 is executed by the processor 16 of the terminal device 10 and the processor 23 of the server 20 . Steps S33 to S35 in FIG. 10 are the same as steps S15 to S17 in FIG. 6, so description thereof will be omitted.

In the data collection process according to the present embodiment, as shown in FIG. It is determined whether or not (step S31). Specifically, whether or not the user has suffered from a specific disease is input by the user via the input device 13, and the determination unit 166 determines whether the user has suffered from the specific disease based on the information input by the user. Determine whether or not you have.

When it is determined whether or not the user has a specific disease, the data control unit 167 of the terminal device 10, based on the determination result, similarly to step S13 in FIG. A target data amount of each parameter to be transmitted to the server 20 per unit time is set (step S32). The data control unit 167 of the terminal device 10 may set the target data amount for each parameter, or may collectively set the target data amount for each of all the parameters. The target data amount is set with the maximum value of the data amount that can be acquired by the terminal device 10 from the sensor 12 or the like as the upper limit.

In particular, the data control unit 167 controls the data amount of each parameter acquired and transmitted to the server 20 when the terminal device 10 is held by a user who has had a specific disease. The amount of data is controlled so as to be greater than the amount of data of each parameter acquired and transmitted to the server 20 when held by an object that has never experienced an event.

In this embodiment, the terminal device 10 determines whether or not the user has a specific disease and sets the target data amount. Therefore, the amount of data transfer between the terminal device 10 and the server 20 can be reduced.

Although preferred embodiments according to the present invention have been described above, the present invention is not limited to these embodiments, and various modifications and changes can be made within the scope of the claims.

1 machine learning system 4 communication network 5 wireless base station 10 terminal device 20 server

Claims (7)

A data collection device for collecting, from a terminal device, parameter data used for learning a machine learning model that outputs output parameters relating to the occurrence of a specific event about an object,
a storage device for storing data;
a determination unit that determines whether or not the specific event has occurred in the object holding the terminal device;
a data control unit that controls the amount of data of each parameter acquired by the terminal device and stored in the storage device per unit time;
The data control unit controls the amount of data of each parameter acquired by the terminal device held by the object in which the specific event has occurred and stored in the storage device so that the amount of data of each parameter obtained by the specific event has occurred. A data collection device that controls the amount of data so that the amount of data is greater than the amount of data of each parameter that is acquired by the terminal device held by the object and stored in the storage device. The data control unit adjusts the data to be stored in the storage device among the data of each parameter transmitted from the terminal device to the data collection device, so that the data acquired by the terminal device and stored in the storage device 2. The data collection device according to claim 1, which controls the amount of data for each parameter. The data control unit adjusts the data transmitted to the data collection device among the data of each parameter detected by the terminal device, thereby adjusting each parameter acquired by the terminal device and stored in the storage device. 3. The data collection device according to claim 1 or 2, which controls the amount of data in . The data control unit adjusts the data acquired from the detector among the data of each parameter detectable by the detector of the terminal device, thereby adjusting each parameter acquired by the terminal device and stored in the storage device. 4. The data collection device according to any one of claims 1 to 3, which controls the data amount of the parameter. said output parameter is the probability of occurrence of said particular event;
When the probability calculated in the past is relatively high, the data control unit acquires by the terminal device and stores in the storage device more than when the probability calculated in the past is relatively low. 5. The data collection device according to any one of claims 1 to 4, wherein said data amount is controlled so that the data amount of each parameter to be collected increases. A data collection method for collecting, from a terminal device, parameter data used for learning a machine learning model that outputs output parameters relating to the occurrence of a specific event about an object,
Determining whether the specific event has occurred in an object holding the terminal device;
controlling the amount of data of each parameter acquired by the terminal device and stored in the storage device of the data collection device per unit time;
The amount of data of each parameter acquired by the terminal device held by the object in which the specific event has occurred and stored in the storage device is obtained by the object in which the specific event has never occurred A data collection method in which the amount of data of each parameter obtained by the held terminal device is larger than the amount of data stored in the storage device. A terminal device that transmits parameter data used for learning a machine learning model that outputs output parameters related to the occurrence of a specific event about an object to a data collection device,
a determination unit that determines whether or not the specific event has occurred in the object holding the terminal device;
a data control unit that controls the amount of data of each parameter transmitted from the terminal device to the data collection device per unit time;
The data control unit controls the amount of data of each parameter to be acquired and transmitted to the data collection device when the terminal device is held by an object in which the specific event has occurred. A terminal device that controls the amount of data to be larger than the amount of data of each parameter acquired and transmitted to the data collection device when held by an object in which the specific event has never occurred. .

Images