JP7483827B1 - Learning device, learning method, and learning program - Google Patents

Abstract

The present invention aims to improve the accuracy of a model by using learning data suitable for sequential learning using the JIT method in imitation learning.
[Solution] The processing device 10 has a collection unit 132 that collects first data which is a combination of explanatory variables and objective variables, an acquisition unit 133 that acquires second data from the first data based on at least the distance between the explanatory variables and a specified explanatory variable, a narrowing down unit 134 that narrows down learning data from the second data based on specified narrowing down conditions, and an update unit 135 that uses the learning data to update a model that outputs the objective variable from the explanatory variables.
[Selected figure] Figure 2

Description

The present invention relates to a learning device, a learning method, and a learning program.

A technique known as imitation learning is known in which a machine learning model learns human behavior and then uses the model to teach actions to humans or robots, etc.

There is also a known technique known as the Just-In-Time (JIT) method, which accumulates a large amount of observed data, extracts data near the desired point from the accumulated data, and sequentially trains a model using the extracted data (see, for example, Non-Patent Document 1).

Here, for example, in a chemical plant, environmental changes occur over time, such as deterioration of equipment with age, deterioration of catalysts, and changes to production load plans.

In response to this, it is possible to apply the JIT method to imitation learning, which learns how operators operate equipment in chemical plants, and adapt the model to changes in the environment.

JP 2019-185194 A

Shigeru Yamamoto, "Just-In-Time Predictive Control: Predictive Control Based on Accumulated Data," Instrumentation and Control, Vol. 52, No. 10, October 2013 (https://www.jstage.jst.go.jp/article/sicejl/52/10/52_878/_pdf/-char/ja)

However, the operator's operation history includes history that is not suitable for learning, such as unsteady states and driving by inexperienced operators. On the other hand, the operator's operation history also includes history that is actively desired for the model to learn, such as driving by experienced operators. If all such operator operation history is learned uniformly, the accuracy of the model may not improve.

If a person were to individually select the history they want the model to learn and the history they do not want the model to learn, there is a risk that the learning will be arbitrary, as the person making the selection will be subjective. Furthermore, in this case, there is the problem that a cumbersome processing burden will be required to manually select huge amounts of data on operation history on a regular basis.

The present invention has been made in consideration of the above, and aims to provide a learning device, a learning method, and a learning program that can improve the accuracy of a model by using learning data suitable for sequential learning using the JIT method in imitation learning.

To solve the above-mentioned problems and achieve the objective, the learning device is characterized by having a collection unit that collects first data, which is a combination of explanatory variables and objective variables, an acquisition unit that acquires second data from the first data based on at least the distance between the explanatory variables and a specified explanatory variable, a narrowing unit that narrows down learning data from the second data based on a predetermined narrowing condition, and an update unit that uses the learning data to update a model that outputs the objective variable from the explanatory variables.

According to the present invention, learning data suitable for sequential learning using the JIT method is used in imitation learning to improve the accuracy of the model.

FIG. 1 is a diagram illustrating a plant operation system. FIG. 2 is a diagram illustrating an example of a configuration of a processing device according to an embodiment. FIG. 3 is a diagram illustrating an example of the history DB. FIG. 4 is a diagram for explaining the processing of the processing device. FIG. 5 is a flowchart showing a processing procedure according to the embodiment. FIG. 6 is a flowchart showing a processing procedure according to the embodiment. FIG. 7 is a diagram illustrating an example of a computer that executes a program.

Below, embodiments of the learning device, learning method, and learning program according to the present application will be described in detail with reference to the drawings. Note that the present invention is not limited to the embodiments described below.

[Embodiment]
[Configuration of the embodiment]
First, a plant operation system will be described with reference to Fig. 1. The plant operation system 1 is a system for managing and controlling a production process of a product in a plant. The plant includes a chemical plant for producing chemical products.

As shown in FIG. 1, the system has a processing device 10, a terminal device 20, and a plant system 30.

The processing device 10 performs processing related to a model for performing imitation learning. The processing device 10 can function as a learning device.

The processing device 10 and the plant system 30 are also connected to each other so that they can communicate data with each other via a network. For example, the network is the Internet and an intranet.

The plant system 30 may include equipment used in the production process and a distributed control system (DCS). For example, the equipment may be a reactor, a cooler, a gas-liquid separator, etc.

The terminal device 20 is an information processing device such as a personal computer, a tablet terminal, or a smartphone, or a dedicated terminal for operating plant equipment.

The operator is a user who operates the equipment included in the plant system 30 via the terminal device 20. The models used in the processing device 10 are managed appropriately by a system administrator, etc.

The processing of each device in the plant operation system 1 will be explained based on Figure 1.

The terminal device 20 operates the equipment of the plant system 30 in response to the operation of the operator (step S1). For example, the terminal device 20 sets the temperature within the equipment, the pressure within the equipment, the target production volume in the production process, the amount of raw materials to be input into the equipment, etc., through the operation.

The plant system 30 operates according to the operation from the terminal device 20 (step S2). Then, the plant system 30 transmits the operation history to the processing device 10 (step S3).

For example, the history includes sensor values of sensors installed at various locations in the plant system 30 and setting values set by operations from the terminal device 20. The history may also be time-series data in which each record is assigned a time (time stamp).

The terminal device 20 transmits the narrowing down conditions for the history to be used as learning data in response to the operation of the operator (step S4). The narrowing down conditions are preset so as to narrow down the history corresponding to the identification information of the operator who instructed the operation, a specified period, a specified situation, and/or the contents of a specified operation. The narrowing down conditions may also be set by a system administrator, etc.

The processing device 10 narrows down the learning data to be used for learning the model from the history, assigns weights, for example, learns the model, and performs inference using the model (step S5). Details of each process of the processing device 10 will be described later.

The processing device 10 then provides the inference result to the operator's terminal device 20 (step S6). For example, the inference result is an operation content predicted from the situation, and a guidance screen 21 showing the predicted operation content is displayed on the screen of the terminal device 20. The operator operates the plant system 30, for example, in accordance with the operation content of the guidance screen 21 displayed on the terminal device 20 (step S1).

The model used for inference learns the operations of the operator through imitation learning. Therefore, other operators can imitate the operations by following the operations obtained as a result of inference by the model.

The processing device 10 will be described in detail with reference to FIG. 2. FIG. 2 is a diagram showing an example of the configuration of the processing device 10 according to an embodiment.

As shown in FIG. 2, the processing device 10 has a communication unit 11, a memory unit 12, and a control unit 13.

The communication unit 11 communicates data with other devices via a network. For example, the communication unit 11 is a network interface card (NIC).

The storage unit 12 is a storage device such as a hard disk drive (HDD) or a solid state drive (SSD). Note that the storage unit 12 may also be a semiconductor memory in which data can be rewritten, such as a random access memory (RAM), a flash memory, or a non-volatile static random access memory (NVSRAM).

The memory unit 12 stores the OS (Operating System) and various programs executed by the processing device 10. The memory unit 12 stores a history DB 121, a filtering condition DB 122, and model information 123.

History DB121 is information including history provided from plant system 30. FIG. 3 is a diagram showing an example of history DB121. As shown in FIG. 3, history DB121 includes a list of explanatory variables such as time, operator, situation, and operation, and a set value which is an objective variable. History DB121 may also include weights.

The time indicates the time when the operation was performed. The operator item indicates identification information of the operator who operated the equipment. The status item includes items such as a first temperature, a second temperature, and a first flow rate. The status item may include a first pressure, a second pressure, a _CO2 concentration, and the like in addition to the items shown in FIG. 3.

The first temperature, the second temperature, the first pressure, the second pressure, and the first flow rate are sensor values of sensors installed at various locations in the plant system 30.

The first temperature, the second temperature, the first pressure, the second pressure, and the first flow rate are explanatory variables of the model and are examples of explanatory variables that represent the situation in the product production process.

The _CO2 concentration is the concentration of _CO2 generated in the production process.

The time is a time stamp indicating the date and time when the first temperature, the second temperature, the first pressure, the second pressure, the first flow rate, and the _CO2 concentration were acquired.

The operation is, for example, the SV value (set value) and MV value (control output) of PID control, which is a control method often used in plant control, and is a set value set by operation from the terminal device 20. The set value may be a normalized value of the actually set value. The set value corresponds to the objective variable of the model.

The set value is the objective variable of the model and is an example of an objective variable that represents the operation of equipment in the production process.

For example, FIG. 3 shows that, as a result of operations by "Yamada Taro," at the time "13:21:01," the first temperature is "102.1°C," the second temperature is "103.0°C," the first flow rate is "311.2 ^m3 /s," and the operation is "206.3."

The filtering condition DB122 is information including preset filtering conditions. The filtering conditions are conditions for selecting learning data for a model. The filtering conditions are set, for example, by the terminal device 20 in response to an operation by an operator. History searched from the history DB121 based on the filtering conditions is used as learning data for the model.

For example, the filtering conditions may be set to the identification information of a specific operator. In this case, the history resulting from the operations of the operator set by the filtering conditions is narrowed down as learning data. When there are experienced operators and inexperienced operators and it is desired to have the model imitate the history of the experienced operator, it is possible to narrow down only the history resulting from operations of this experienced operator by setting the identification information of the experienced operator as the filtering conditions. Note that the history resulting from operations by an experienced team leader, etc. may be weighted more heavily than other histories.

In addition, the filtering conditions are set to a time or a period. For example, if the daytime is a busy time and active driving is possible, but the nighttime is a lighter time and active driving decreases, the filtering conditions can be set to a daytime period (e.g., 8:00 to 18:00), making it possible to filter out only the daytime history. Note that the history for a specified period during the day may be weighted.

In addition, a situation is set as the filtering condition. For example, a situation in which the average fluctuation between the first temperature and the second temperature is less than a threshold is set as the filtering condition. When it is desired to exclude history relating to unusual driving from the learning data, a situation in which unusual driving situations are excluded is set as the filtering condition, making it possible to exclude history relating to unusual driving from the learning data.

Examples of unusual operations include factory start-up and shutdown, temporary maintenance work, etc. During such operations, for example, the average of the first temperature and the second temperature may increase by 1 degree or more compared to 30 minutes prior, and the filtering conditions are set so that the history for 30 minutes from such operations is excluded. Specifically, the filtering conditions are set to be the average of the first temperature and the second temperature, and a situation in which the increase from the average 30 minutes prior is less than 1 degree Celsius.

In addition, the narrowing condition is set to a situation where the variance of the target value for operation over a specified period of time falls below a threshold. If you want the model to mimic the history of stable quality, you can set the narrowing condition to a situation where the variance of the target value for operation over the past 30 minutes falls below 0.3, for example. This makes it possible to narrow down the history of cases where quality has stabilized through operation.

In addition, the implementation content is set as the narrowing down condition. For example, the implementation content is set to be when the setting value of a specific device is changed and then returned to the setting value before the change within a specific period of time. For example, in order to use the history of the case where the valve opening is changed and then returned to the original setting value within 5 minutes as learning data, the narrowing down condition is set to be the case where the valve opening setting is changed and then returned to the original valve opening setting other than 5 minutes after the change.

The model information 123 is information such as parameters for constructing a model. If the model is a neural network, the model information 123 is the weights and biases of each layer, and if the model is Lasso regression or Ridge regression, it is regularization parameters. Furthermore, the model information 123 includes parameters such as the order of preprocessing and the window width (window size) in moving average processing.

The control unit 13 controls the entire processing device 10. The control unit 13 is, for example, an electronic circuit such as a CPU (Central Processing Unit), MPU (Micro Processing Unit), or GPU (Graphics Processing Unit), or an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array).

The control unit 13 also has an internal memory for storing programs and control data that define various processing procedures, and executes each process using the internal memory. The control unit 13 also functions as various processing units by the operation of various programs. For example, the control unit 13 has a reception unit 131, a collection unit 132, an acquisition unit 133, a narrowing down unit 134, an update unit 135, an inference unit 136, and a display control unit 137. The processing of each functional unit of the processing device 10 will be explained using Figure 4. Figure 4 is a diagram explaining the processing of the processing device 10.

The reception unit 131 receives, for example, the filtering conditions from the terminal device 20 ((1) in FIG. 4) and stores them in the filtering condition DB 122. The reception unit 131 not only receives new filtering conditions, but also receives modifications and deletions (including partial deletions).

The collection unit 132 collects the operation history of the plant system 30 ((2) in FIG. 4) and stores the collected history in the history DB 121. The history is an example of first data, which is a combination of explanatory variables and objective variables.

The acquisition unit 133 acquires history from the history included in the history DB 121 based on the distance between the explanatory variable and the specified explanatory variable. The history included in the history DB 121 is an example of first data. The history acquired by the acquisition unit 133 is an example of second data. The acquisition unit 133 may acquire second data from the data (first data) included in the history DB 121 based on the distance between the explanatory variable and the specified explanatory variable and the weight.

When a history search key (explanatory variables and/or objective variables) is specified, the acquisition unit 133 acquires a group of past histories similar to this history search key from the history DB 121 ((3) in FIG. 4). The acquisition unit 133 acquires from the history DB 121 a group of past histories (explanatory variables, objective variables) (second data) collected in a situation similar to the history search key (e.g., explanatory variables indicating the current situation).

The specified explanatory variable is called a request point. For example, a request point is an explanatory variable (corresponding to each sensor value in history DB121) at a specific time. Note that the objective variable (set value) at the request point may be unknown.

Here, in the JIT method, similar history is acquired based on the distance between the training data (corresponding to the history DB 121 in this embodiment), which is a multidimensional vector, and the request point, which is also a multidimensional vector. For example, the acquisition unit 133 uses the JIT method to acquire k Nearest Neighbors (k-NN), which are k (k is an integer) records with small calculated Euclidean distances. Note that the distance between the training data and the request point is not limited to the Euclidean distance, and may be, for example, the Mahalanobis distance or the cosine similarity.

The acquisition unit 133 may acquire records by referring to the weights in the history DB 121, in addition to the distance between the training data and the request point. Here, if the larger the weight, the more desirable the data is to be acquired, for example, the acquisition unit 133 preferentially acquires data from the history DB 121 that is k-nearest and has a large weight.

The narrowing down unit 134 narrows down the learning data from the past history group (second data) acquired by the acquisition unit 133 based on a predetermined narrowing down condition ((4) in FIG. 4). As described above, the narrowing down condition is a condition that is preset to narrow down the past history group to data corresponding to the identification information of the operator who instructed the operation, a predetermined period, a predetermined situation, and/or a setting value set by the operation, and is read from the narrowing down condition DB 122.

When the operator's identification information is set in the narrowing down condition, the narrowing down unit 134 narrows down the second data to the history of the operations of the operator set in the narrowing down condition as learning data.

When a time or period is set as the filtering condition, the filtering unit 134 narrows down the history of the time or period set as the filtering condition from the second data as learning data.

When a situation is set as a filtering condition, the filtering unit 134 filters the history corresponding to the situation set as the filtering condition from the second data as learning data.

When implementation details are set as the filtering conditions, the filtering unit 134 filters the history corresponding to the implementation details set as the filtering conditions from the second data as learning data.

The acquisition unit 133 may use a specified history search key (explanatory variables and/or objective variables) and a filtering condition to acquire a group of past histories that are similar to the history search key and match the filtering condition from the history DB 121. In this case, the filtering unit 134 is omitted. In this way, the acquisition unit 133 can reduce bias in the number of learning data by filtering the group of past histories in a single process.

The update unit 135 uses the learning data narrowed down by the narrowing down unit 134 to learn a model that outputs a target variable from an explanatory variable ((5) in FIG. 4), and updates the model ((6) in FIG. 4).

The update unit 135 calculates an objective function that represents the difference between the objective variables calculated by inputting explanatory variables into the model constructed from the model information 123 and the objective variables included in the learning data narrowed down by the narrowing down unit 134, and repeatedly updates the model parameters, i.e., the model information 123, until the learning termination condition is satisfied so that the objective function becomes smaller. Note that if weights are assigned to the learning data, the learning data is learned with the assigned weights.

The inference unit 136 calculates the objective variable by inputting the explanatory variables for prediction into the model constructed from the updated model information 123. That is, the inference unit 136 performs inference processing ((7) in FIG. 4). The inferred objective variable is, for example, the operation content predicted from the situation.

The display control unit 137 presents the inference results to the operator by displaying a guidance screen showing the inferred objective variable (e.g., operation content) on the terminal device 20.

[Processing of the embodiment]
The processing procedure in the embodiment will be described with reference to Fig. 5. Fig. 5 is a flow chart showing the processing procedure in the embodiment.

As shown in FIG. 5, first, the processing device 10 accepts the narrowing down conditions, for example, from the terminal device 20 (step S11). The processing device 10 is not limited to accepting new narrowing down conditions, and may accept additions, modifications, deletions, etc. of conditions. Then, the processing device 10 stores the accepted narrowing down conditions in the DB 122 (step S12).

Figure 6 is a flowchart showing the processing procedure in the embodiment. As shown in Figure 6, the processing device 10 collects the operation history of the plant system 30 (step S21) and stores the collected history in the history DB 121.

The processing device 10 extracts data from the history contained in the history DB 121 based on the distance between the explanatory variable and the specified explanatory variable. When a history search key (explanatory variable and/or objective variable) is specified, the processing device 10 retrieves a group of past histories similar to this history inspection key from the history DB 121 (step S22).

The processing device 10 narrows down the learning data from the history group acquired in step S22 based on the filtering conditions stored in the filtering condition DB 122 (step S23).

The processing device 10 uses the learning data narrowed down by the narrowing down unit 134 to learn a model that outputs a target variable from an explanatory variable (step S24), and updates the model (step S25).

The processing device 10 infers the objective variable (e.g., operation content) by inputting explanatory variables for prediction (e.g., the first temperature, the second temperature, the first flow rate, etc.) into the model constructed from the updated model information 123 (step S26).

The processing device 10 presents the inference result to the operator by displaying a guidance screen on the terminal device 20 that shows the objective variable (e.g., the operation content) inferred in step S25 (step S27).

[Effects of the embodiment]
In this way, the processing device 10 according to the embodiment collects first data, which is a combination of explanatory variables and objective variables, and obtains second data from the first data based on at least the distance between the explanatory variables and the specified explanatory variables. The processing device 10 then narrows down the learning data from the second data based on a predetermined narrowing condition, and uses the learning data to update a model that outputs objective variables from explanatory variables.

In this way, the processing device 10 automatically narrows down a huge amount of history data to history suitable for learning according to the filtering conditions and updates the model, eliminating the need for manual selection of history.

The processing device 10 also narrows down the learning data suitable for learning from the history collected from the plant system 30 according to specified filtering conditions, so that it can properly select history suitable for learning without the inclusion of the subjectivity of the person selecting it. The processing device 10 can also accurately narrow down the history suitable for learning using filtering conditions set to correspond to the identification information of the operator who instructed the operation, a specified period, a specified situation, and/or a setting value set by the operation.

The processing device 10 can therefore update the model using only history suitable for learning, thereby improving the inference accuracy of the model. In particular, the processing device 10 updates the model using learning data suitable for sequential learning using the JIT method in imitation learning, thereby enabling appropriate operation of the plant system 30 and improved model accuracy to be achieved in parallel.

[System configuration, etc.]
In addition, each component of each device shown in the figure is functionally conceptual, and does not necessarily have to be physically configured as shown in the figure. In other words, the specific form of distribution and integration of each device is not limited to that shown in the figure, and all or a part of it can be functionally or physically distributed or integrated in any unit depending on various loads, usage conditions, etc. Furthermore, each processing function performed by each device can be realized in whole or in any part by a CPU (Central Processing Unit) and a program analyzed and executed by the CPU, or can be realized as hardware by wired logic. Note that the program may be executed not only by the CPU but also by other processors such as a GPU.

Furthermore, among the processes described in this embodiment, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using known methods. In addition, the information including the processing procedures, control procedures, specific names, various data, and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified.

[program]
In one embodiment, the processing device 10 can be implemented by installing a learning program that executes the above learning process as package software or online software on a desired computer. For example, the above learning program can be executed by an information processing device, causing the information processing device to function as the processing device 10. The information processing device here includes desktop or notebook personal computers. In addition, the information processing device also includes tablet terminals, smartphones, mobile communication terminals such as mobile phones and PHS (Personal Handyphone System), and slate terminals such as PDA (Personal Digital Assistant).

The processing device 10 can also be implemented as a server that provides services related to the above-mentioned learning process to a client, the client being a terminal device used by a user. For example, the server is implemented as a server device that provides a learning service in which the specification of a request point is used as input and a trained model is used as output. In this case, the server may be implemented as a web server, or may be implemented as a cloud that provides services related to the above-mentioned learning process by outsourcing.

Figure 7 is a diagram showing an example of a computer that executes a program. The computer 1000 has, for example, a memory 1010 and a CPU 1020. The computer 1000 also has a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. Each of these components is connected by a bus 1080.

The memory 1010 includes a read only memory (ROM) 1011 and a random access memory (RAM) 1012. The ROM 1011 stores a boot program such as a basic input output system (BIOS). The hard disk drive interface 1030 is connected to a hard disk drive 1090. The disk drive interface 1040 is connected to a disk drive 1100. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1100. The serial port interface 1050 is connected to a mouse 1110 and a keyboard 1120, for example. The video adapter 1060 is connected to a display 1130, for example.

The hard disk drive 1090 stores, for example, an OS 1091, an application program 1092, a program module 1093, and program data 1094. That is, the program that defines each process of the processing device 10 is implemented as a program module 1093 in which computer-executable code is written. The program module 1093 is stored, for example, in the hard disk drive 1090. For example, a program module 1093 for executing processes similar to the functional configuration of the processing device 10 is stored in the hard disk drive 1090. The hard disk drive 1090 may be replaced by an SSD (Solid State Drive).

The setting data used in the processing of the above-described embodiment is stored as program data 1094, for example, in memory 1010 or hard disk drive 1090. Then, the CPU 1020 reads the program module 1093 or program data 1094 stored in memory 1010 or hard disk drive 1090 into RAM 1012 as necessary, and executes the processing of the above-described embodiment.

The program module 1093 and the program data 1094 are not limited to being stored in the hard disk drive 1090, but may be stored in, for example, a removable storage medium and read by the CPU 1020 via the disk drive 1100 or the like. Alternatively, the program module 1093 and the program data 1094 may be stored in another computer connected via a network (such as a local area network (LAN) or wide area network (WAN)). The program module 1093 and the program data 1094 may then be read by the CPU 1020 from the other computer via the network interface 1070.

Reference Signs List 1 Plant operation system 10 Processing device 20 Terminal device 30 Plant system 11 Communication unit 12 Storage unit 13 Control unit 121 History DB
122 Narrowing Condition DB
123 Model information 131 Reception unit 132 Collection unit 133 Acquisition unit 134 Narrowing unit 135 Update unit 136 Inference unit 137 Display control unit

Claims (5)

A collection unit that collects first data, which is a combination of explanatory variables and objective variables;
an acquisition unit that acquires second data from the first data based on at least a distance between the explanatory variable and a specified explanatory variable;
a narrowing-down unit that narrows down learning data from the second data based on a predetermined narrowing-down condition;
an update unit that updates a model that outputs the objective variable from the explanatory variables using the learning data;
having
As the narrowing condition, a situation is set such that an unusual driving situation is excluded,
The narrowing-down unit is configured to exclude history data relating to unusual driving from the learning data based on the narrowing-down conditions. the first data is a combination of the explanatory variables representing a situation in a production process of a product and the objective variables representing an operation of a device in the production process;
The learning device according to claim 1, characterized in that the narrowing-down conditions are conditions that are preset to narrow down the second data to data corresponding to the identification information of the operator who instructed the operation, a specified period, a specified situation, and/or a setting value set by the operation. The learning device according to claim 2, characterized in that the specified situation is when the average fluctuation between the first temperature and the second temperature is less than a threshold value and/or when the variance of the value that is the operation target for a specified period of time is below a threshold value. A learning method executed by a learning device, comprising:
collecting first data which is a combination of explanatory variables and objective variables;
acquiring second data from the first data based at least on a distance between the explanatory variable and a specified explanatory variable;
Narrowing down learning data from the second data based on a predetermined narrowing down condition;
updating a model that outputs the objective variable from the explanatory variables using the learning data;
Including,
As the narrowing condition, a situation is set such that an unusual driving situation is excluded,
A learning method characterized in that the narrowing down step excludes the history of driving that is different from normal from the learning data based on the narrowing down conditions. A step of collecting first data which is a combination of explanatory variables and a response variable;
acquiring second data from the first data based at least on a distance between the explanatory variable and a specified explanatory variable;
Narrowing down learning data from the second data based on a predetermined narrowing down condition;
updating a model that outputs the objective variable from the explanatory variables using the learning data;
Run the following on your computer:
As the narrowing condition, a situation is set such that an unusual driving situation is excluded,
The narrowing down step is a learning program for excluding the history of driving that is different from the normal driving from the learning data based on the narrowing down conditions.

Images