Method, information processing device, and program

Abstract

To provide a technology for more accurately predicting the probability of achieving a target according to the blending conditions of raw materials. [Solution] The method includes the steps of inputting input data including raw material blending conditions into a trained classification model and obtaining the probability that a material produced under those blending conditions will be classified into each of multiple evaluation results specified for a predetermined first evaluation test; predicting the probability of achieving a first target, that is, that the material will satisfy the desired first conditions, based on the multiple probabilities obtained in the obtaining step; inputting the input data into a trained regression model and obtaining, for the material, a probability distribution of predicted values ​​for the physical properties of the material to be evaluated in a second evaluation test different from the first evaluation test; predicting the probability of achieving a second target, that is, that the material will satisfy the desired second conditions, based on the probability distribution obtained in the obtaining step; and integrating the probability of achieving the first target and the probability of achieving the second target.

Description

[Technical Field]

[0001] The present disclosure relates to a method, an information processing device, and a program. [Background technology]

[0002] Japanese Patent Laid-Open Publication No. 2023-126824 (Patent Document 1) discloses a composition search method for more efficiently searching for a composition for obtaining a target physical property value.

[0003] The composition search method generates a prediction model by learning training data in which information about the composition of a material is used as an explanatory variable and the physical property values ​​of the material are used as target variables. The composition search method then inputs prediction data indicating the composition into the prediction model to calculate predicted values ​​of the physical properties. The composition search method then calculates the distance between the prediction data and the training data and weights the distance with a weight corresponding to the influence of each explanatory variable on the prediction. The composition search method then displays the relationship between the predicted value and the weighted distance. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2023-126824 Summary of the Invention [Problem to be solved by the invention]

[0005] Incidentally, there is a need to know the probability with which a material will achieve a target condition by changing the compounding conditions of the composition, raw materials, etc. In this regard, the composition search method disclosed in Patent Document 1 predicts the Young's modulus from the composition of the material (paragraph

[0101] ), but does not predict the probability of achieving the target.

[0006] The present disclosure has been made to solve the above-mentioned problems, and an object of one aspect is to provide a technology for more accurately predicting the probability of target achievement according to raw material blending conditions. [Means for solving the problem]

[0007] An example of the present disclosure provides a method executed by an information processing device, comprising the steps of: inputting input data including blending conditions of raw materials into a trained classification model and acquiring a probability that a material produced under the blending conditions will be classified into each of a plurality of evaluation results specified for a predetermined first evaluation test; predicting a first target achievement probability that the material will satisfy a desired first condition based on the plurality of probabilities acquired in the acquiring step; inputting the input data into a trained regression model and acquiring, for the material, a probability distribution of predicted values ​​of a physical property of an evaluation target in a second evaluation test different from the first evaluation test; predicting a second target achievement probability that the material will satisfy a desired second condition based on the probability distribution acquired in the acquiring step; and integrating the first target achievement probability and the second target achievement probability.

[0008] In one example of the present disclosure, the classification model is generated by a learning process using a plurality of performance data, each of which associates blending conditions of raw materials with evaluation results of the first evaluation test performed on materials produced under the blending conditions.

[0009] In one example of the present disclosure, the regression model is generated by a learning process using the plurality of performance data, each of which associates blending conditions of raw materials with the evaluation results of the first evaluation test conducted on the material produced under the blending conditions, and the evaluation results of the second evaluation test conducted on the material produced under the blending conditions.

[0010] In one example of the present disclosure, the regression model is a Gaussian process regression model that is trained to receive input of blending conditions of raw materials and output a probability distribution of predicted values ​​for the physical properties.

[0011] In one example of the present disclosure, the first evaluation test is a qualitative evaluation test, and the second evaluation test is a quantitative evaluation test.

[0012] In one example of the present disclosure, in the integrating step, the first target achievement probability and the second target achievement probability are integrated for each of the plurality of input data for which different blending conditions are specified. The method includes a step of outputting input data from among the plurality of input data that maximizes the integration result in the integrating step.

[0013] In one example of the present disclosure, the blending conditions include at least one of the blending ratio of the raw materials and the manufacturing conditions when blending the raw materials.

[0014] In one example of the present disclosure, the material comprises a mucoadhesive material.

[0015] In one example of the present disclosure, the method further includes a step of calculating, for each of the plurality of performance data, a similarity between a combination condition defined in the performance data and a combination condition defined in the input data, and a step of weighting the first target achievement probability with a weight corresponding to the plurality of similarities calculated in the calculating step. In the integrating step, the first target achievement probability after weighting in the weighting step and the second target achievement probability are integrated.

[0016] In another example of the present disclosure, an information processing device is provided. The information processing device includes a control unit. The control unit executes the following steps: inputting input data including raw material blending conditions into a trained classification model and acquiring a probability that a material produced under the blending conditions will be classified into each of multiple evaluation results specified for a predetermined first evaluation test; predicting a first target achievement probability that the material will satisfy a desired first condition based on the multiple probabilities acquired in the acquiring step; inputting the input data into a trained regression model and acquiring, for the material, a probability distribution of predicted values ​​of physical properties of an evaluation target in a second evaluation test different from the first evaluation test; predicting a second target achievement probability that the material will satisfy a desired second condition based on the probability distribution acquired in the acquiring step; and integrating the first target achievement probability and the second target achievement probability.

[0017] Another example of the present disclosure provides a program executed by an information processing device. The program causes the information processing device to execute the following steps: inputting input data including blending conditions of raw materials into a trained classification model and acquiring a probability that a material produced under the blending conditions will be classified into each of multiple evaluation results specified for a predetermined first evaluation test; predicting a first target achievement probability that the material will satisfy a desired first condition based on the multiple probabilities acquired in the acquiring step; inputting the input data into a trained regression model and acquiring, for the material, a probability distribution of predicted values ​​of physical properties of the evaluation target in a second evaluation test different from the first evaluation test; predicting a second target achievement probability that the material will satisfy a desired second condition based on the probability distribution acquired in the acquiring step; and integrating the first target achievement probability and the second target achievement probability.

[0018] The above and other objects, features, aspects and advantages of the present invention will become apparent from the following detailed description of the invention taken in conjunction with the accompanying drawings. [Brief explanation of the drawings]

[0019] [Figure 1]FIG. 2 is a diagram for explaining an outline of functions of an information processing device. [Figure 2] FIG. 2 illustrates an example of a hardware configuration of an information processing device. [Figure 3] FIG. 10 is a diagram illustrating an example of a training dataset. [Figure 4] FIG. 10 is a diagram illustrating an example of a prediction dataset. [Figure 5] FIG. 2 is a diagram illustrating an example of a functional configuration of an information processing apparatus. [Figure 6] FIG. 2 is a diagram conceptually illustrating a learning process performed by a learning unit. [Figure 7] FIG. 10 is a diagram conceptually illustrating a prediction process performed by an achievement probability calculation unit. [Figure 8] FIG. 10 is a diagram showing an output result by an achievement probability calculation unit. [Figure 9] FIG. 10 is a diagram conceptually illustrating a prediction process performed by an achievement probability calculation unit. [Figure 10] FIG. 10 is a diagram conceptually illustrating an integration process performed by an integration unit. [Figure 11] 10 is a flowchart showing the flow of a learning process. [Figure 12] 10 is a flowchart showing the flow of a prediction process. [Figure 13] 13 is a flowchart showing the process flow of step S200A shown in FIG. 12. [Figure 14] 13 is a flowchart showing the process flow of step S200B shown in FIG. 12. [Figure 15] FIG. 10 is a diagram illustrating an outline of a correction process for a target achievement probability. [Figure 16] FIG. 10 illustrates an example of a functional configuration of an information processing device according to a second embodiment. [Figure 17] FIG. 10 is a diagram conceptually illustrating a prediction process according to a second embodiment. [Figure 18] FIG. 10 is a diagram conceptually illustrating weighting processing according to the second embodiment. [Figure 19] FIG. 10 is a diagram conceptually illustrating integration processing according to the second embodiment. [Figure 20]It is a flowchart showing the flow of prediction processing according to the second embodiment. [Figure 21] It is a flowchart showing the flow of the process of step S200C shown in FIG. 20.

Embodiments for Carrying Out the Invention

[0020] Hereinafter, each embodiment according to the present invention will be described with reference to the drawings. In the following description, the same parts and components are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed descriptions thereof will not be repeated. Note that each of the embodiments and each modification described below may be selectively combined as appropriate.

[0021] [First Embodiment] [A. Overview] FIG. 1 is a diagram for explaining an outline of the functions of the information processing apparatus 100 according to the present embodiment.

[0022] The information processing apparatus 100 is, for example, a desktop PC (Personal Computer), a notebook PC, a tablet terminal, a smartphone, or other computer.

[0023] The information processing apparatus 100 has a function of predicting the target achievement probability that the material manufactured under the blending conditions satisfies the desired target conditions from the input data for prediction including the blending conditions of the raw materials. Hereinafter, an outline of the function of predicting the target achievement probability will be described.

[0024] The prediction of the target achievement probability is realized using the learned classification model 126 and the learned regression model 127.

[0025] The classification model 126 is generated in advance by a learning process using a plurality of performance data. Each of the plurality of performance data associates the blending conditions of the raw materials with the evaluation results of a qualitative evaluation test (first evaluation test) performed on the material manufactured under the blending conditions.

[0026] Qualitative evaluation tests are tests that manually evaluate evaluation scales that are difficult to quantify using equipment according to predetermined criteria. In qualitative evaluation tests, evaluation ranks are assigned. In qualitative evaluation tests, the order of the evaluation ranks is meaningful, but the size of the evaluation rank is meaningless. For example, when comparing evaluation ranks "1" and "2" with evaluation ranks "2" and "3," the rank difference is both "1," but the value of the rank difference itself is meaningless. Specific examples of qualitative evaluation tests will be discussed later.

[0027] Similarly, regression model 127 is generated in advance by a learning process using a plurality of pieces of performance data, each of which associates raw material blending conditions with the evaluation results of a quantitative evaluation test (second evaluation test) conducted on a material produced under the blending conditions.

[0028] A quantitative evaluation test is a test that evaluates the physical properties of a material using measuring equipment. That is, in a quantitative evaluation test, the measurement value output from the measuring equipment becomes the evaluation result. In a quantitative evaluation test, the magnitude of the measurement value itself is meaningful. Specific examples of quantitative evaluation tests will be described later.

[0029] For ease of explanation, hereinafter, the evaluation results (e.g., evaluation ranks) of qualitative evaluation tests are also referred to as “qualitative measures,” and the evaluation results (e.g., measured values) of quantitative evaluation tests are also referred to as “quantitative measures.” Specific examples of quantitative evaluation tests will be described later.

[0030] In step S1, the information processing device 100 inputs input data 125 including blending conditions of raw materials to the trained classification model 126. As a result, the classification model 126 outputs the probability (hereinafter also referred to as "classification probability") that a material produced under the blending conditions specified in the input data 125 will be classified into each evaluation result in a qualitative evaluation test.

[0031] Next, in step S2, the information processing device 100 predicts the target achievement probability that the material produced under the blending conditions defined in the input data 125 will satisfy the desired target condition (first condition) based on each classification probability acquired in step S1. As an example, the information processing device 100 sums the classification probabilities for each evaluation rank belonging to the target condition, and predicts the sum as the target achievement probability.

[0032] Next, in step S3, the information processing device 100 inputs the input data 125 to the trained regression model 127. As a result, in step S4, the information processing device 100 acquires a probability distribution of predicted values ​​regarding the physical properties of the evaluation target in the quantitative evaluation test.

[0033] The method for acquiring the probability distribution of the predicted values ​​is arbitrary. As an example, the regression model 127 is a Gaussian process regression model 326. The Gaussian process regression model 326 is trained to receive input of the blending conditions of the raw materials and output a probability distribution regarding the evaluation results of a quantitative evaluation test. The output value of the Gaussian process regression model 326 includes a predicted value and a deviation from the average value of the predicted value.

[0034] As another example, the information processing device 100 uses multiple regression models 127 trained using different learning algorithms. In this case, even if the regression models 127 receive the same input data 125, they output different predicted values ​​for the physical properties of the object to be evaluated in a quantitative evaluation test. The information processing device 100 calculates the average value of the multiple predicted values ​​output from each regression model 127 and calculates the standard deviation corresponding to the multiple predicted values. Thereafter, the information processing device 100 obtains a probability distribution (e.g., a normal distribution) according to the calculated average value and standard deviation.

[0035] The following description is based on the Gaussian process regression model 326, but the regression model 127 is not limited to the Gaussian process regression model 326.

[0036] Next, in step S5, the information processing apparatus 100 integrates the target achievement probability calculated based on the output of the classification model 126 and the target achievement probability calculated based on the output of the Gaussian process regression model 326. As an example, the information processing apparatus 100 integrates the respective target achievement probabilities by multiplying them together.

[0037] Thereby, the information processing apparatus 100 can integrate the evaluation scales of the qualitative evaluation test and the quantitative evaluation test with the same type of scale called the target achievement probability. As a result, the information processing apparatus 100 can specify the blending conditions that satisfy the target conditions with respect to both the physical properties related to the qualitative evaluation test and the physical properties related to the quantitative evaluation test.

[0038] Note that, in the above description, an example using one classification model 126 has been described, but two or more classification models may be used. As an example, N (N is an integer of 2 or more) classification models learned for N qualitative evaluation tests may be used respectively. In this case, the information processing apparatus 100 calculates the target achievement probability for each of the classification models.

[0039] Also, in the above description, an example using one Gaussian process regression model 326 has been described, but two or more Gaussian process regression models may be used. As an example, M (M is an integer of 2 or more) Gaussian process regression models learned for M quantitative evaluation tests may be used respectively. In this case, the information processing apparatus 100 calculates the target achievement probability for each of the Gaussian process regression models.

[0040] The information processing apparatus 100 integrates the respective target achievement probabilities by multiplying the N target achievement probabilities calculated for each of the classification models and the M target achievement probabilities calculated for each of the Gaussian process regression models.

[0041] <B. Hardware Configuration of Information Processing Apparatus 100> Next, the hardware configuration of the information processing device 100 will be described with reference to Fig. 2. Fig. 2 is a diagram showing an example of the hardware configuration of the information processing device 100.

[0042] The information processing device 100 includes a control device 101 (control unit), a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, a communication interface 104, a display interface 105, an input interface 107, and an auxiliary storage device 120. These components are connected to a bus 115.

[0043] The control device 101 is configured, for example, by at least one integrated circuit. The integrated circuit may be configured, for example, by at least one central processing unit (CPU), at least one graphics processing unit (GPU), at least one application specific integrated circuit (ASIC), at least one field programmable gate array (FPGA), or a combination thereof.

[0044] The control device 101 controls the operation of the information processing device 100 by executing various programs. Upon receiving an execution command for one of the programs, the control device 101 reads the program to be executed from the auxiliary storage device 120 or the ROM 102 into the RAM 103. The RAM 103 functions as a working memory and temporarily stores various data required for executing the program.

[0045] A LAN (Local Area Network), an antenna, etc. are connected to the communication interface 104. The information processing device 100 exchanges data with external devices via the communication interface 104. The external devices include, for example, servers.

[0046] A display device 106 is connected to the display interface 105. The display interface 105 sends an image signal for displaying an image to the display device 106 in accordance with a command from the control device 101 or the like. The display device 106 is, for example, a liquid crystal display, an organic EL (Electro Luminescence) display, or other displays. The display device 106 may be configured integrally with the information processing device 100 or may be configured separately from the information processing device 100.

[0047] An input device 108 is connected to the input interface 107. The input device 108 is, for example, a mouse, a keyboard, a touch panel, or any other device capable of receiving user operations. The input device 108 may be configured integrally with the information processing device 100, or may be configured separately from the information processing device 100.

[0048] The auxiliary storage device 120 is a storage medium such as a hard disk or a flash memory. The auxiliary storage device 120 stores, for example, the above-mentioned training dataset 122, the above-mentioned prediction dataset 124 including the input data 125, the above-mentioned classification model 126, the training program 128, the prediction program 130, the above-mentioned Gaussian process regression model 326, the training program 328, and the prediction program 330. These may be stored in a storage location other than the auxiliary storage device 120, such as a storage area (e.g., cache memory) of the control device 101, the ROM 102, the RAM 103, an external device (e.g., a server), or the like.

[0049] The learning program 128 is a program for generating the classification model 126 using the learning dataset 122. The learning program 128 may be provided not as a standalone program but as part of an arbitrary program. In this case, the learning process by the learning program 128 is realized in cooperation with the arbitrary program. Even a program that does not include some of these modules does not deviate from the spirit of the learning program 128 according to this embodiment. Furthermore, some or all of the functions provided by the learning program 128 may be realized by dedicated hardware. Furthermore, the information processing device 100 may be configured in the form of a so-called cloud service in which at least one server executes part of the processing of the learning program 128.

[0050] The prediction program 130 is a program for predicting the goal achievement probability according to input data using the classification model 126. The prediction program 130 may be provided not as a standalone program but as part of an arbitrary program. In this case, the learning process by the prediction program 130 is realized in cooperation with the arbitrary program. Even a program that does not include some of these modules does not deviate from the spirit of the prediction program 130 according to this embodiment. Furthermore, some or all of the functions provided by the prediction program 130 may be realized by dedicated hardware. Furthermore, the information processing device 100 may be configured in the form of a so-called cloud service in which at least one server executes part of the processing of the prediction program 130.

[0051] The learning program 328 is a program for generating the Gaussian process regression model 326 using the learning dataset 122. The learning program 328 may be provided not as a single program but incorporated into a part of any program. In this case, the learning process by the learning program 328 is realized in cooperation with any program. Even a program that does not include such partial modules does not deviate from the gist of the learning program 328 according to the present embodiment. Further, part or all of the functions provided by the learning program 328 may be realized by dedicated hardware. Further, the information processing apparatus 100 may be configured in a form such as a so-called cloud service in which at least one server executes part of the processing of the learning program 328.

[0052] The prediction program 330 is a program for predicting the goal achievement probability according to the input data using the Gaussian process regression model 326. The prediction program 330 may be provided not as a single program but incorporated into a part of any program. In this case, the learning process by the prediction program 330 is realized in cooperation with any program. Even a program that does not include such partial modules does not deviate from the gist of the prediction program 330 according to the present embodiment. Further, part or all of the functions provided by the prediction program 330 may be realized by dedicated hardware. Further, the information processing apparatus 100 may be configured in a form such as a so-called cloud service in which at least one server executes part of the processing of the prediction program 330.

[0053] <C. Learning Dataset 122> Next, referring to FIG. 3, the learning dataset 122 shown in FIGS. 1 and 2 will be described. FIG. 3 is a diagram showing an example of the learning dataset 122.

[0054] As described above, the information processing apparatus 100 generates the classification model 126 and the Gaussian process regression model 326 by learning the learning dataset 122.

[0055] The training data set 122 includes a plurality of pieces of performance data 123. The number of pieces of performance data 123 included in the training data set 122 is arbitrary. As an example, the number of pieces of performance data 123 ranges from several to tens of thousands.

[0056] More specifically, a data ID (Identification) is defined for each piece of performance data 123. The data ID is an identifier for uniquely identifying the performance data 123. The data ID is defined by, for example, a combination of letters, numbers, and symbols.

[0057] Each piece of performance data 123 associates an explanatory variable with a response variable. The values ​​of the explanatory variables defined in the performance data 123 and the values ​​of the response variables defined in the performance data 123 are actually measured values.

[0058] The explanatory variables defined in the performance data 123 are, for example, the blending conditions of the raw materials. The blending conditions include, for example, at least one of the blending ratio of each type of raw material and the manufacturing conditions when manufacturing the material from each raw material. Examples of the manufacturing conditions include the temperature during manufacturing, the pressure applied during manufacturing, and the manufacturing time.

[0059] The objective variables defined in the performance data 123 include the evaluation rank (evaluation result) given in the qualitative evaluation test and the measurement value related to the quantitative evaluation test. That is, each piece of performance data 123 associates the blending conditions of the raw materials, the evaluation result of the qualitative evaluation test conducted on the material produced under those blending conditions, and the evaluation result of the quantitative evaluation test conducted on the material produced under those blending conditions.

[0060] As mentioned above, qualitative evaluation tests are tests that manually evaluate evaluation scales that are difficult to quantify using equipment according to predetermined criteria. For example, an evaluator visually inspects a material and assigns a label to the material based on a predetermined evaluation standard, indicating the material's physical properties. The evaluation rank may be indicated by two values, "OK" and "NG," or by three or more values.

[0061] Materials to be evaluated include, for example, adhesive materials and inks. "Adhesive materials" refers to materials that have the function of joining objects together. Examples of adhesive materials include adhesive tapes and adhesives.

[0062] An example of a qualitative evaluation test is a test for evaluating the adhesiveness of cellophane tape. In this evaluation test, whether or not a surface coating, paint film, print, etc. peels off when an adhesive tape attached to an object is peeled off is evaluated. The evaluator evaluates the adhesiveness of cellophane tape, for example, using a rating scale of "1" to "5." A rating scale of "5" indicates "no peeling." A rating scale of "4" indicates "slight peeling." A rating scale of "3" indicates "partial peeling." A rating scale of "2" indicates "considerable peeling." A rating scale of "1" indicates "almost complete peeling."

[0063] Another example of a qualitative evaluation test is a test for evaluating offset. In this evaluation test, it is evaluated whether ink applied to a sheet-like object or an adhesive material attached to a sheet-like object will penetrate to the back side or be transferred to another object. The evaluator evaluates the offset using an evaluation rank of "1" to "5," for example. As an example, an evaluation rank of "5" indicates "no offset." An evaluation rank of "4" indicates "almost no offset." An evaluation rank of "3" indicates "slight offset." An evaluation rank of "2" indicates "clear offset." An evaluation rank of "1" indicates "significant offset."

[0064] Another example of a qualitative evaluation test is an appearance evaluation test. Examples of appearance evaluation tests include a test to evaluate the presence or absence of scratches, a test to evaluate the presence or absence of cracks, a test to evaluate the presence or absence of foreign matter, a test to evaluate color, and a test to evaluate gloss. In an appearance evaluation test, an evaluator assigns an evaluation rank based on the appearance of the material.

[0065] The qualitative evaluation test is not limited to the visual evaluation test described above, but may be an evaluation test in which there is a maximum value in the evaluation criteria. Examples of such evaluation tests include a friction resistance test, a peel strength test, and a strength test of a laminating adhesive.

[0066] In a rub resistance test, an evaluator rubs a sample (such as ink) applied to a printed material under certain conditions and checks for changes in the sample's surface condition. The evaluator then records the number of rubs at which the surface deteriorates as an evaluation value. The upper limit of the rub count is, for example, 200 rubs. In other words, if the surface condition does not change after the evaluator rubs 200 times, the product is deemed to have passed. In this case, all evaluation values ​​for passing products are the same, which serves as a qualitative evaluation index. The evaluator evaluates the rub resistance using, for example, an evaluation rank of "0" or "1." More specifically, if the sample does not deteriorate even after the upper limit of rub counts, the evaluator assigns an evaluation rank of "1," indicating a pass. On the other hand, if the sample deteriorates before the upper limit of rub counts is reached, the evaluator assigns an evaluation rank of "0," indicating a fail.

[0067] In a peel strength test, the adhesive strength of a sample is evaluated, in which a substrate such as a film is bonded with an adhesive or the like. The evaluator evaluates the adhesive strength as the strength required to peel the adhesive material from the substrate. The adhesive strength is measured using a measuring device. The evaluator evaluates the peel strength using, for example, an evaluation rank of "0" or "1." More specifically, if the measured adhesive strength meets the pass criterion, the evaluator assigns an evaluation rank of "1," indicating a pass. On the other hand, if the substrate peels off before the adhesive strength reaches the upper limit, the evaluator assigns an evaluation rank of "0," indicating a fail.

[0068] In a quantitative evaluation test, the performance of a material as an evaluation result is expressed on a quantitative scale using a measuring instrument. As an example, the quantitative scale as an explanatory variable includes at least one of the strength required to peel the adhesive material from the adherend (hereinafter also referred to as "adhesion strength"), the resistance of the adhesive material to slippage from the adherend (hereinafter also referred to as "holding strength"), and the ease of adhesion of the adhesive material to the adherend (hereinafter also referred to as "tack").

[0069] Examples of adhesive strength include peel adhesive strength, shear adhesive strength, and split adhesive strength.

[0070] The peel adhesive strength is expressed, for example, as the force required to peel the adhesive material in a direction perpendicular to the adherend surface (so-called 90° peel adhesive strength), or as the force required to peel the adhesive material in a direction parallel to the adherend surface (so-called 180° peel adhesive strength).

[0071] The shear adhesive strength corresponds to the force required to break the bonded portion when forces (i.e., shear stress) are applied in opposite directions in a horizontal plane to adherends bonded together by an adhesive material.

[0072] The splitting adhesive strength corresponds to the force exerted when adherends joined together by an adhesive material are peeled apart when forces are applied in opposite directions on a vertical plane to the adherends.

[0073] Examples of holding power include shear holding power and constant load holding power. Shear holding power is expressed as the distance an adhesive material displaces when a load is applied in the shear direction to the adhesive material attached to an adherend for a certain period of time. Alternatively, holding power is expressed as the time from when a load is applied in the shear direction to the adhesive material attached to an adherend until the adhesive material falls. On the other hand, constant load holding power is expressed as the distance an adhesive material displaces when a load is applied in the perpendicular direction to the adhesive material attached to an adherend, or the time from when the load is applied until the adhesive material falls.

[0074] Examples of tack include ball tack. Ball tack is measured by rolling balls of different diameters in sequence on an inclined surface to which an adhesive material is attached. The ball tack is represented by the maximum diameter of the ball that stops on the inclined surface.

[0075] Note that the quantitative scale (explanatory variable) defined in the performance data 123 is not limited to the above examples. As another example, the quantitative scale may be the viscosity of the material measured by a measuring instrument. As still another example, the quantitative scale may be the friction resistance (e.g., coefficient of friction, number of friction cycles, etc.) measured by a measuring instrument in a friction resistance test such as a rubbing test. As still another example, the quantitative scale may be the adhesive strength measured by a measuring instrument for the laminating adhesive.

[0076] <D. Prediction data set 124> Next, referring to FIG. 4, the prediction data set 124 shown in FIG. 2 will be described. FIG. 4 is a diagram showing an example of the prediction data set 124.

[0077] The prediction data set 124 includes a plurality of input data 125. The number of input data 125 included in the prediction data set 124 is arbitrary. As an example, the number of input data 125 is several to tens of thousands.

[0078] Each of the input data 125 is defined with a data ID. The data ID is an identifier for uniquely identifying the input data 125. The data ID is defined, for example, by a combination of characters, numerical values, and symbols.

[0079] Each of the input data 125 defines the blending conditions of the raw materials as explanatory variables. The types of explanatory variables defined in the input data 125 are the same as those defined in the above-mentioned performance data 123 (see FIG. 3). More specifically, the blending conditions defined in the input data 125 include, for example, at least one of the blending ratios of the respective raw materials and the manufacturing conditions when manufacturing the material from each raw material. Examples of the manufacturing conditions include the temperature during manufacturing, the pressure applied during manufacturing, and the manufacturing time.

[0080] The blending ratio defined for each of the input data 125 may be randomly determined within the range specified by the evaluator, or may be arbitrarily specified by the evaluator.

[0081] <E. Functional Configuration of Information Processing Apparatus 100> Next, referring to FIGS. 5 to 10, the functional configuration of the information processing apparatus 100 will be described. FIG. 5 is a diagram showing an example of the functional configuration of the information processing apparatus 100.

[0082] As shown in FIG. 5, the information processing apparatus 100 includes, as functional components, a learning unit 152, an achievement probability calculation unit 154, a learning unit 352, an achievement probability calculation unit 354, an integration unit 359, and an output unit 360. Hereinafter, these functional components will be described in order.

[0083] Note that it is not necessary for all of the learning unit 152, the achievement probability calculation unit 154, the learning unit 352, the achievement probability calculation unit 354, the integration unit 359, and the output unit 360 to be implemented in the information processing apparatus 100. Some of the functional components may be implemented in the information processing apparatus 100, and the remaining functional components may be implemented in another computer such as a server.

[0084] As an example, the achievement probability calculation unit 154, the achievement probability calculation unit 354, the integration unit 359, and the output unit 360 may be implemented in the information processing apparatus 100, and the learning unit 152 and the learning unit 352 may be implemented in another computer.

[0085] (E1. Learning Unit 152) First, the function of the learning unit 152 shown in Fig. 5 will be described with reference to Fig. 6. Fig. 6 is a diagram conceptually showing the learning process performed by the learning unit 152.

[0086] The learning unit 152 executes a learning process using the above-described learning dataset 122 composed of a plurality of pieces of performance data 123, and generates a classification model 126. The machine learning algorithm employed is not particularly limited, and various machine learning algorithms such as a neural network such as deep learning, a support vector machine, or a decision tree system may be employed. The learning process using a neural network will be described below.

[0087] The classification model 126 is composed of an input layer X, a hidden layer H, and an output layer Y.

[0088] The input layer X is configured to receive input of explanatory variables defined in the performance data 123. As an example, the input layer X includes a unit group x1 and a unit group x2.

[0089] The unit group x1 is configured to receive explanatory variables (e.g., raw material blending ratios) defined in the performance data 123. The number of units constituting the unit group x1 is the same as the number of dimensions of the blending ratios. As an example, if the blending ratios are five-dimensional, the unit group x1 is composed of five units. Each unit constituting the unit group x1 outputs the input data to each unit in the first layer of the intermediate layer H.

[0090] The unit group x2 is configured to receive input of other explanatory variables (e.g., manufacturing conditions) defined in the classification model 126. The number of units constituting the unit group x2 is the same as the number of dimensions of the manufacturing conditions. As an example, if the manufacturing conditions are expressed in two dimensions, the unit group x2 is composed of two units. Each unit constituting the unit group x2 outputs the input data to each unit in the first layer of the intermediate layer H.

[0091] The middle layer H is composed of multiple layers. The number of layers in the middle layer H is arbitrary. In the example of FIG. 6, the middle layer H is composed of N layers (N is a natural number). Each layer in the middle layer H includes multiple units. In the example of FIG. 6, the first layer of the middle layer H is composed of unit h A1 ,h A2 The final layer of the intermediate layer H is made up of units h N1 ,h N2 It is composed of...

[0092] Each unit constituting each layer of the hidden layer H is connected to each unit in the previous layer and each unit in the next layer. Each unit in each layer receives each output value from each unit in the previous layer, multiplies each output value by a weight, accumulates the multiplication results, adds (or subtracts) a predetermined bias to (or from) the accumulation result, inputs the addition result (or subtraction result) to a predetermined function (for example, a sigmoid function), and outputs the output value of the function to each unit in the next layer.

[0093] The output layer Y outputs an evaluation result (i.e., classification probability) according to the explanatory variables input to the input layer X. As an example, the output layer Y is composed of units y1 to y3. Hereinafter, the units y1 to y3 will also be referred to as unit y.

[0094] Each unit y is connected to each unit h in the final layer of the hidden layer H. N1 ,h N2 Each of the units y receives an output value from each unit in the final layer of the hidden layer H, multiplies each output value by a weight, accumulates the results of these multiplications, adds (or subtracts) a predetermined bias to (or from) the accumulated result, inputs the result of the addition (or subtraction) to a predetermined function (for example, a sigmoid function), and outputs the output result of the function as an output value.

[0095] The number of units constituting the output layer Y is determined according to the number of objective variables defined in the performance data 123. As described above, the objective variables are evaluation ranks given in a qualitative evaluation test, and the number of units constituting the output layer Y is, for example, the same as the number of evaluation ranks.

[0096] Unit y1 is configured to output a classification probability belonging to evaluation rank "1." Unit y2 is configured to output a classification probability belonging to evaluation rank "2." Unit y3 is configured to output a classification probability belonging to evaluation rank "3."

[0097] Note that the output value of the classification model 126 does not need to be a probability itself, but may be a value correlated with the probability. As an example, the output value of the classification model 126 may be a score correlated with the classification probability. In this case, the information processing device 100 normalizes the scores so that the sum of the scores becomes 1, and calculates the normalized score as the classification probability.

[0098] Furthermore, in the example of FIG. 6, one classification model 126 is configured to output a classification probability belonging to evaluation ranks "1" to "3", but one classification model 126 may output one probability. As an example, the first classification model 126 is configured to receive input of the blending conditions of the ingredients and output a probability belonging to evaluation rank "1". The second classification model 126 is configured to receive input of the blending conditions of the ingredients and output a probability belonging to evaluation rank "2". The third classification model 126 is configured to receive input of the blending conditions of the ingredients and output a probability belonging to evaluation rank "3".

[0099] Next, the update process of the internal parameters of the classification model 126 by the learning unit 152 will be described.

[0100] The learning unit 152 acquires the first piece of performance data 123 from the training dataset 122, and inputs the explanatory variables defined in the performance data 123 to the classification model 126. As a result, the classification model 126 outputs the probability of belonging to each evaluation rank. Next, the learning unit 152 compares the output prediction results "p1" to "p3" with the objective variable defined in the first piece of performance data 123. The objective variable is expressed, for example, as probabilities "YA" to "YC".

[0101] As an example, when the evaluation rank associated with the performance data 123 is "1", the value of the objective variable is (YA, YB, YC) = (1, 0, 0). When the evaluation rank associated with the performance data 123 is "2", the value of the objective variable is (YA, YB, YC) = (0, 1, 0). When the evaluation rank associated with the performance data 123 is "3", the value of the objective variable is (YA, YB, YC) = (0, 0, 1).

[0102] The learning unit 152 calculates the error "Z" between the output results "p1" to "p3" of the classification model 126 and the objective variables "YA" to "YC." As an example, the error "Z" is calculated based on the following formula (1).

[0103] Z={(p1-YA) 2 +(p2-YB) 2 +(p3-YC) 2} / 3···(1) Next, the learning unit 152 updates various parameters (for example, weights and biases) included in the classification model 126 so as to reduce the error "Z." The parameter update is realized, for example, by the backpropagation algorithm.

[0104] The learning unit 152 repeatedly updates the internal parameters of the classification model 126 for each piece of performance data 123 included in the learning dataset 122. As a result, the classification model 126 begins to output accurate prediction results as the learning progresses.

[0105] Note that the learning unit 152 does not need to use all of the performance data 123 included in the training data set 122 for the learning process, and may generate the classification model 126 using a portion of the performance data 123 included in the training data set 122. The remaining performance data 123 is used, for example, to evaluate the classification model 126.

[0106] (E2. Achievement Probability Calculation Unit 154) Next, the function of the achievement probability calculation unit 154 shown in Fig. 5 will be described with reference to Fig. 7. Fig. 7 is a diagram conceptually showing the prediction process by the achievement probability calculation unit 154.

[0107] The achievement probability calculation unit 154 receives input of target conditions 132 related to the physical properties of the material. The target conditions 132 are input, for example, to an input screen displayed on the above-mentioned display device 106. The evaluator inputs the target conditions 132 to the input screen using, for example, the above-mentioned input device 108.

[0108] The target condition 132 is a condition for specifying a range of evaluation ranks associated with the output layer of the classification model 126. In this case, the target condition 132 is defined by at least one threshold value.

[0109] Next, the achievement probability calculation unit 154 sequentially acquires the input data 125 included in the above-mentioned prediction data set 124 (see FIG. 4 ), and inputs the input data 125 to the classification model 126. As a result, the classification model 126 outputs the probability of being classified into each evaluation rank defined for the qualitative evaluation test.

[0110] When a numerical range equal to or less than a predetermined threshold is specified as the goal condition 132, the achievement probability calculation unit 154 sums up the classification probabilities for the evaluation ranks equal to or less than the predetermined threshold, and calculates the sum result as the goal achievement probability "P pred Here, "i" is the number of the input data 125 defined in the prediction dataset 124. As an example, when the input goal condition 132 indicates an evaluation rank of "2" or less, the achievement probability calculation unit 154 calculates the goal achievement probability "P (i)" as the sum of the classification probability "p1" for the evaluation rank "1" and the classification probability "p2" for the evaluation rank "2". pred (i)" is calculated as follows.

[0111] When a numerical range equal to or greater than a predetermined threshold is specified as the goal condition 132, the achievement probability calculation unit 154 sums up the classification probabilities for evaluation ranks equal to or greater than the threshold, and calculates the sum as the goal achievement probability "P predAs an example, when the input goal condition 132 indicates an evaluation rank of "2" or higher, the achievement probability calculation unit 154 calculates the goal achievement probability "P (i)" as the sum of the classification probability "p2" for the evaluation rank "2" and the classification probability "p3" for the evaluation rank "3." pred (i)" is calculated as follows.

[0112] When a numerical range equal to or greater than the first threshold and equal to or less than the second threshold is specified as the goal condition 132, the achievement probability calculation unit 154 sums up the classification probabilities for the evaluation ranks that are equal to or greater than the first threshold and equal to or less than the second threshold, and calculates the sum result as the goal achievement probability "P pred (i)" is calculated as follows.

[0113] 8 is a diagram showing an output result RS1A by the achievement probability calculation unit 154. As shown in FIG. 8, the output result RS1A is calculated for each piece of input data 125 input to the classification model 126. The output result RS1A is output to the integration unit 359.

[0114] (E3. Learning Section 352) Next, the function of the learning unit 352 shown in Fig. 5 will be described with reference to Fig. 3 above. The learning unit 352 executes a learning process using the learning dataset 122 described above to generate a Gaussian process regression model 326. The machine learning algorithm employed is not particularly limited. As an example, a Bayesian optimization algorithm or the like may be employed.

[0115] Gaussian process regression model 326 is trained to receive input of raw material blending conditions and output a probability distribution regarding the evaluation results of a quantitative evaluation test. The output value of Gaussian process regression model 326 includes a predicted value and a deviation from the predicted value as a mean value.

[0116] The learning unit 352 estimates hyperparameters of the kernel function using the given training dataset 122. As an example, the learning unit 352 performs maximum likelihood estimation by maximizing the marginal log-likelihood of the Gaussian process. The marginal log-likelihood is calculated by Gaussian integration of the distribution of observed data. The hyperparameters are efficiently estimated by using an optimization method such as gradient descent. The Gaussian process regression model 326 predicts the mean and deviation of the predicted value from the posterior distribution of the Gaussian process.

[0117] For example, an API (Application Programming Interface) provided by an individual or a company is used as a learning program for generating the Gaussian process regression model 326. Examples of such APIs include libraries available as open source in Python and libraries released by Google as open source (e.g., TensorFlow).

[0118] (E4. Achievement Probability Calculation Unit 354) Next, the function of the achievement probability calculation unit 354 shown in Fig. 5 will be described with reference to Fig. 9. Fig. 9 is a diagram conceptually showing the prediction process by the achievement probability calculation unit 354.

[0119] The achievement probability calculation unit 354 receives input of desired target conditions 332 related to the physical properties of the material. The target conditions 332 are input, for example, to an input screen displayed on the display device 106 described above. The evaluator inputs the target conditions 332 to the input screen using, for example, the input device 108 described above. The target conditions 332 are conditions for specifying a range in the probability distribution output from the Gaussian process regression model 326. Note that the target conditions 332 are different from the target conditions 132 described above.

[0120] Next, the achievement probability calculation unit 354 acquires the above-mentioned prediction data set 124 (see FIG. 4 ) and sequentially inputs the input data 125 included in the prediction data set 124 to the Gaussian process regression model 326. As a result, the Gaussian process regression model 326 outputs a probability distribution regarding the physical property evaluated in the quantitative evaluation test.

[0121] In Fig. 9, the probability distribution of the predicted value for the blending conditions is shown as graph G1. For ease of explanation, the explanatory variables are shown as one dimension in Fig. 9, but in reality, the explanatory variables have two or more dimensions.

[0122] As an example, the Gaussian process regression model 326 receives input of "XA" as a blending condition, and outputs a predicted value "YA" and a deviation "σ" based on "YA". As a result, the achievement probability calculation unit 354 acquires a probability distribution NR. The probability distribution NR is, for example, a normal distribution with the predicted value "YA" as the mean value and the deviation "σ" as the standard deviation.

[0123] As an example, assume that a numerical range equal to or less than a predetermined threshold is specified as the target condition 332. In this case, the achievement probability calculation unit 354 integrates the probability density in the range equal to or less than the predetermined value in a probability density function (PDF) according to the probability distribution NR, and calculates the integration result as the target achievement probability.

[0124] As another example, suppose a numerical range equal to or greater than a predetermined threshold is specified as the target condition 332. In this case, the achievement probability calculation unit 354 integrates the probability densities in the range equal to or greater than the predetermined value in a probability density function according to the probability distribution NR, and calculates the integration result as the target achievement probability.

[0125] As yet another example, suppose a numerical range equal to or greater than a first threshold and equal to or less than a second threshold is specified as the goal condition 332. In this case, the achievement probability calculation unit 354 integrates the probability densities in the range equal to or greater than the first threshold and equal to or less than the second threshold in a probability density function according to the probability distribution NR, and calculates the integration result as the goal achievement probability.

[0126] (E5. Integration Section 359) Next, the function of the integrating unit 359 shown in Fig. 5 will be described with reference to Fig. 10. Fig. 10 is a diagram conceptually showing the integrating process performed by the integrating unit 359.

[0127] 10 shows the output result RS1A from the achievement probability calculation unit 154 and the output result RS1B from the achievement probability calculation unit 354. The integrating unit 359 integrates the output result RS1A from the achievement probability calculation unit 154 and the output result RS1B from the achievement probability calculation unit 354.

[0128] As an example, the integration unit 359 may use the target achievement probability "P pred (i)" and the target achievement probability "P''" specified in the output result RS1B. pred (i)" to calculate the integrated result RS5.

[0129] (E6. Output section 360) Continuing with reference to FIG. 10, the function of the output unit 360 shown in FIG. 5 will be described.

[0130] The output unit 360 outputs the input data 125 that is a candidate for a prescription based on the integration result RS5 by the integration unit 359.

[0131] As an example, the output unit 360 may refer to the integration result RS5 and output the maximum goal achievement probability "P calc (i) and determine the target achievement probability "P calc The input data 125 corresponding to (i) is output as a prescription candidate.

[0132] As another example, the output unit 360 may refer to the integration result RS5 and output the target achievement probability "P calc (i) and determine the target achievement probability "P calc The input data 125 corresponding to (i) is output as a prescription candidate.

[0133] The method of outputting the prescription candidate is arbitrary. As an example, the output unit 360 displays the blending conditions defined in the input data 125 selected as the prescription candidate on the display device 106 described above. As another example, the output unit 360 stores the blending conditions defined in the input data 125 selected as the prescription candidate in the auxiliary storage device 120 described above.

[0134] <Flowchart related to learning process> Next, referring to FIG. 11, the flow of the learning process by the information processing apparatus 100 will be described. FIG. 11 is a flowchart showing the flow of the learning process.

[0135] The control device 101 of the information processing apparatus 100 functions as the learning unit 152 described above by executing the learning program 128 described above, and executes the learning process shown in FIG. 11. In other aspects, part or all of the learning process may be executed by circuit elements or other hardware.

[0136] In step S110, the control device 101 initializes the variable "j". As an example, the variable "j" is initialized to "1".

[0137] In step S112, the control device 101 acquires the j-th performance data 123 included in the learning data set 122.

[0138] In step S114, the control device 101 inputs the blending conditions as explanatory variables defined in the performance data 123 acquired in step S112 into the classification model 126. As a result, the classification model 126 outputs the probabilities of being classified into each evaluation label defined for the qualitative evaluation test.

[0139] In step S116, the control device 101 calculates the error between the classification probability as the correct value defined in each performance data 123 obtained in step S112 and the classification probability as the prediction result obtained in step S114, and updates the internal parameters of the classification model 126 so that the error becomes smaller than the current error. The parameters are updated, for example, by the error backpropagation method.

[0140] In step S120, the control device 101 determines whether to end the learning process. As an example, when the variable "j" is greater than a predetermined value, the control device 101 determines to end the learning process. Alternatively, when all the performance data 123 included in the learning data set 122 have been used for learning, the control device 101 determines to end the learning process. When the control device 101 determines to end the learning process (YES in step S120), it ends the process shown in FIG. 11. Otherwise (NO in step S120), the control device 101 switches the control to step S122.

[0141] In step S122, the control device 101 increments the variable "j". That is, the control device 101 increases the variable "j" by 1. After that, the control device 101 returns the control to step S112.

[0142] <G. Flowchart related to prediction processing> Next, referring to FIG. 12, the flow of the prediction process by the information processing device 100 will be described. FIG. 12 is a flowchart showing the flow of the prediction process.

[0143] The control device 101 of the information processing device 100 executes the prediction process shown in FIGS. 12 and 13 by executing the above-described prediction programs 130 and 330. In other aspects, part or all of the prediction process may be executed by circuit elements or other hardware.

[0144] In step S200, the control device 101 initializes a variable “n.” As an example, the variable “n” is initialized to “1.”

[0145] In step S202, the control device 101 acquires the type of physical property related to the n-th prediction target. Note that the type of physical property of the prediction target and the prediction order are set in advance by, for example, an evaluator.

[0146] In step S204, the control device 101 determines whether the prediction target acquired in step S202 is a qualitative scale. If the control device 101 determines that the type of the prediction target is a qualitative scale (YES in step S204), it switches control to step S200A. If not (NO in step S204), the control device 101 switches control to step S200B.

[0147] In step S200A, the control device 101 uses the above-described classification model 126 to calculate the target achievement probability according to the input data 125. That is, in step S200A, a predicted value of the target achievement probability related to the qualitative evaluation test is obtained. Details of the processing of step S200A will be described later. After executing the processing of step S200A, the control device 101 switches control to step S330.

[0148] In step S200B, the control device 101 uses the above-described Gaussian process regression model 326 to calculate the target achievement probability corresponding to the input data 125. That is, in step S200B, a predicted value of the target achievement probability related to the quantitative evaluation test is obtained. Details of the processing of step S200B will be described later. After executing the processing of step S200B, the control device 101 switches control to step S330.

[0149] In step S330, the control device 101 determines whether prediction processing has been completed for all specified physical properties. If the control device 101 determines that prediction processing has been completed for all specified physical properties (YES in step S330), the control device 101 switches control to step S340. If not (NO in step S330), the control device 101 switches control to step S332.

[0150] In step S332, the control device 101 increments the variable "n." That is, the control device 101 increases the variable "n" by 1. Thereafter, the control device 101 returns the control to step S202.

[0151] In step S340, the control device 101 functions as the above-mentioned integrating unit 359, and integrates the target achievement probabilities obtained in steps S200A and S200B. The integration process is as described above, and therefore description thereof will not be repeated.

[0152] In step S342, the control device 101 functions as the above-mentioned output unit 360, and outputs input data 125 that are prescription candidates based on the target achievement probability after integration calculated in step S340. The prescription candidate output process is as described above, and therefore the description thereof will not be repeated.

[0153] (G1. Step S200A) Next, the process of step S200A shown in Fig. 12 will be described in detail with reference to Fig. 13. Fig. 13 is a flowchart showing the process of step S200A shown in Fig. 12.

[0154] In step S210, the control device 101 initializes a variable "i." As an example, the variable "i" is initialized to "1."

[0155] In step S212, the control device 101 functions as the achievement probability calculation unit 154 described above, and acquires the target condition 132 set for the physical property of the n-th prediction target. The target condition 132 is set in advance by, for example, a user. The target condition 132 is as described above, and therefore will not be described again.

[0156] In step S214, the control device 101 functions as the achievement probability calculation unit 154 described above, and acquires the i-th input data 125 included in the prediction dataset 124. Next, the control device 101 inputs the acquired input data 125 to the classification model 126. As a result, the classification model 126 outputs the probability of being classified into each evaluation rank defined for the qualitative evaluation test.

[0157] In step S216, the control device 101 functions as the achievement probability calculation unit 154 described above, and calculates the target achievement probability "P pred (i) is calculated. The probability of achieving the target, P pred The calculation method for (i) is as described above, and therefore the explanation will not be repeated.

[0158] In step S220, the control device 101 calculates the target achievement probability "P pred As an example, when the variable "i" is greater than a predetermined value, the control device 101 determines whether or not to end the calculation process of the target achievement probability "P pred Alternatively, the control device 101 determines that the calculation process of "(i)" is to be terminated when all of the input data 125 included in the prediction data set 124 has been used in the calculation process. The control device 101 determines that the calculation process is to be terminated when all of the input data 125 included in the prediction data set 124 has been used in the calculation process. pred If it is determined that the calculation process of "(i)" is to be ended (YES in step S220), the process of step S200A is terminated and control is switched to the above-mentioned S330. Otherwise (NO in step S220), the control device 101 switches control to step S222.

[0159] In step S222, the control device 101 increments the variable "i." That is, the control device 101 increases the variable "i" by 1. Thereafter, the control device 101 returns the control to step S214.

[0160] (G2. Step S200B) Next, the process of step S200B shown in Fig. 12 will be described in detail with reference to Fig. 14. Fig. 14 is a flowchart showing the process of step S200B shown in Fig. 12.

[0161] In step S310, the control device 101 initializes a variable "i." As an example, the variable "i" is initialized to "1."

[0162] In step S312, the control device 101 functions as the achievement probability calculation unit 354 described above, and acquires the target condition 332 set for the physical property of the n-th prediction target. The target condition 332 is set in advance by, for example, a user. The target condition 332 is as described above, and therefore will not be described again.

[0163] In step S314, the control device 101 functions as the achievement probability calculation unit 354 described above, and acquires the i-th input data 125 included in the prediction dataset 124. Next, the control device 101 inputs the acquired input data 125 to the Gaussian process regression model 326. As a result, the control device 101 acquires, as a probability distribution, a predicted value corresponding to the input data 125 and a standard deviation with the predicted value as the mean.

[0164] In step S316, the control device 101 functions as the achievement probability calculation unit 354 described above, and calculates the goal achievement probability "P" that satisfies the goal condition 332 acquired in step S312 based on the probability distribution acquired in step S314. pred (i) is calculated. The probability of achieving the target is "P'' pred The calculation method for (i) is as described above, and therefore the explanation will not be repeated.

[0165] In step S320, the control device 101 determines whether to end the calculation process of the target achievement probability "P'' pred (i)". As an example, when the variable "i" is greater than a predetermined value, the control device 101 determines to end the calculation process of the target achievement probability "P'' pred (i)". Alternatively, when the control device 101 has used all the input data 125 included in the prediction data set 124 for the calculation process, the control device 101 determines to end the calculation process. When the control device 101 determines to end the calculation process of the target achievement probability "P'' pred (i)" (YES in step S320), it exits the process of step S200B and switches the control to S330 described above. Otherwise (NO in step S320), the control device 101 switches the control to step S322.

[0166] In step S322, the control device 101 increments the variable "i". That is, the control device 101 increases the variable "i" by 1. After that, the control device 101 returns the control to step S314.

[0167] [Second Embodiment] <H. Overview> Next, the prediction process of the target achievement probability according to the second embodiment will be described.

[0168] The information processing device 100 according to the first embodiment integrated the target achievement probability calculated based on the output of the classification model 126 and the target achievement probability calculated based on the output of the Gaussian process regression model 326. In contrast, the information processing device 100 according to the second embodiment corrects the target achievement probability calculated based on the output of the classification model 126. After that, the information processing device 100 integrates the corrected target achievement probability and the target achievement probability calculated based on the output of the Gaussian process regression model 326.

[0169] An overview of the process of correcting the goal achievement probability will be described below with reference to Fig. 15. Fig. 15 is a diagram showing an outline of the process of correcting the goal achievement probability.

[0170] In step S11, the information processing device 100 inputs input data 125 including blending conditions of raw materials to the classification model 126. As a result, the classification model 126 outputs the probability (i.e., classification probability) that a material produced under the blending conditions specified in the input data 125 will be classified into each evaluation result in a qualitative evaluation test.

[0171] Next, in step S12, the information processing device 100 predicts the target achievement probability that the material manufactured under the blending conditions specified in the input data 125 will satisfy the desired target conditions, based on the classification probabilities acquired in step S11. As an example, the information processing device 100 sums the classification probabilities for each evaluation rank belonging to the target conditions, and predicts the sum as the target achievement probability.

[0172] Next, in step S13, the information processing device 100 calculates, for each of the multiple performance data 123, the similarity between the blending conditions defined in the performance data 123 and the blending conditions defined in the input data 125. The similarity indicates how similar the input data 125 for prediction is to the learning dataset 122.

[0173] Any algorithm may be used to calculate the similarity, such as kernel density estimation (KDE), mean squared error (MSE), sum of squared difference (SSD), sum of absolute difference (SAD), normalized cross-correlation (NCC), or zero-mean normalized cross-correlation (ZNCC).

[0174] The magnitude of the calculated similarity varies depending on the algorithm used. That is, the calculated similarity value may be larger as the performance data 123 and the input data 125 are more similar to each other, or may be smaller as the performance data 123 and the input data 125 are more similar to each other.

[0175] Hereinafter, for ease of understanding, the similarity will be expressed as "distance." That is, a short distance between the performance data 123 and the input data 125 indicates that the performance data 123 and the input data 125 are similar. On the other hand, a long distance between the performance data 123 and the input data 125 indicates that the performance data 123 and the input data 125 are dissimilar.

[0176] Next, in step S14, the information processing device 100 weights the target achievement probability predicted in step S13 with weights (first weights) corresponding to the distances calculated in step S13. In this way, the information processing device 100 corrects the target achievement probability according to the distance from the performance data 123 used for learning. As a result, the information processing device 100 can more accurately predict the target achievement probability according to the blending conditions of the ingredients.

[0177] Typically, the information processing device 100 calculates the weighted target achievement probability for a plurality of input data 125. Then, the information processing device 100 outputs the input data 125 with the highest weighted target achievement probability as a prescription candidate. Alternatively, the information processing device 100 outputs the input data 125 with the weighted target achievement probability that falls within a predetermined top rank as a prescription candidate. This allows the evaluator to efficiently search for raw material blending conditions for a material that satisfies the desired target conditions.

[0178] Preferably, the weight according to the distance is calculated using the distance (hereinafter also referred to as the "minimum distance") relating to the actual data 123 that is most similar to the input data 125 among the actual data 123 included in the learning dataset 122.

[0179] In a certain situation, the information processing apparatus 100 corrects such that the smaller the value of the minimum distance, the smaller the target achievement probability. In other words, the information processing apparatus 100 corrects such that the larger the value of the minimum distance, the larger the target achievement probability. As a result, when there is no performance data 123 similar to the input data 125, the target achievement probability becomes larger. Thereby, it becomes easier for the input data 125 for which the accuracy of the prediction result cannot be compensated to remain as a prescription candidate, and the optimal prescription can be left to the evaluator.

[0180] In another situation, the information processing apparatus 100 corrects such that the smaller the value of the minimum distance, the larger the target achievement probability. In other words, the information processing apparatus 100 corrects such that the larger the value of the minimum distance, the smaller the target achievement probability. As a result, when there is performance data 123 similar to the input data 125, the target achievement probability becomes larger. Thereby, it becomes easier for the input data 125 similar to the performance data 123 to remain as a prescription candidate.

[0181] The information processing apparatus 100 according to the present embodiment integrates the weighted target achievement probability calculated from the output of the classification model 126 and the target achievement probability calculated from the output of the Gaussian process regression model 326.

[0182] <I. Functional Configuration> Next, referring to FIGS. 16 to 19, the functional configuration of the information processing apparatus 100 according to the second embodiment will be described. FIG. 16 is a diagram showing an example of the functional configuration of the information processing apparatus 100 according to the second embodiment.

[0183] As shown in FIG. 16, the information processing apparatus 100 includes, as functional components, a learning unit 152, an achievement probability calculation unit 154, a distance weight calculation unit 156, a correction unit 158, a learning unit 352, an achievement probability calculation unit 354, an integration unit 359, and an output unit 360.

[0184] The functions of learning unit 152, achievement probability calculation unit 154, learning unit 352, achievement probability calculation unit 354, and output unit 360 are as described above, and therefore will not be described repeatedly. Below, the functional configurations of distance weight calculation unit 156, correction unit 158, and integration unit 359 will be described in order.

[0185] 16 does not necessarily have to be implemented in the information processing device 100. A part of the functional configuration shown in Fig. 16 may be implemented in the information processing device 100, and the remaining functional configuration may be implemented in another computer such as a server.

[0186] (I1. Distance weight calculation unit 156) First, the function of distance weight calculation unit 156 shown in Fig. 16 will be described with reference to Fig. 17. Fig. 17 is a diagram conceptually showing the prediction process by distance weight calculation unit 156.

[0187] The distance weight calculation unit 156 calculates, for each of the input data 125 to be predicted, the distance between the combination conditions defined in the input data 125 and the combination conditions defined in the performance data 123. The method for calculating the distance is as described above, and therefore the description thereof will not be repeated.

[0188] Next, the distance weight calculation unit 156 identifies the minimum distance between the actual data 123 and the input data 125 for each of the input data 125 to be predicted. Hereinafter, the minimum distance identified for each of the input data 125 is referred to as "d0(i)".

[0189] 17 shows the output result RS2 by the distance weight calculation unit 156. The output result RS2 shows the minimum distance "d0(i)" identified for each piece of input data 125 to be predicted.

[0190] Next, the distance weight calculation unit 156 normalizes the minimum distance "d0(i)" so that it is equal to or greater than 0 and equal to or less than 1. More specifically, the distance weight calculation unit 156 first calculates the minimum distance "d minNext, the distance weight calculation unit 156 determines the maximum distance "d max Next, the distance weight calculation unit 156 calculates the normalized distance "d(i)" based on the following equation (2).

[0191] d(i)=(d0(i)-d min ) / (d max -d min )···(2)

[0192] (I2. Correction unit 158) Next, the function of the correction unit 158 ​​shown in Fig. 16 will be described with reference to Fig. 18. Fig. 18 is a diagram conceptually showing the weighting process performed by the correction unit 158.

[0193] 18 shows the output result RS1 from the achievement probability calculation unit 154, the output result RS2 from the distance weight calculation unit 156, and the output result RS3 from the correction unit 158. The correction unit 158 ​​corrects the goal achievement probability "P" calculated by the achievement probability calculation unit 154 using a weight according to the distance "d(i)" calculated by the distance weight calculation unit 156. pred As an example, the correction unit 158 ​​corrects the weighted goal achievement probability "P calc (i)" is calculated.

[0194] P calc (i)=P pred (i)·d(i)···(3) In addition, the target achievement probability "P calc The calculation formula for "d(i)" is not limited to the above formula (3). As another example, the correction unit 158 ​​may calculate the goal achievement probability "P calc (i)" is further weighted with a weight "w2" (second weight). Then, the correction unit 158 ​​calculates the weighted goal achievement probability "P pred (i)” and the unweighted goal achievement probability “P pred (i)” and the sum of them is the target achievement probability “P calc (i)” can be used. In this case, the target achievement probability “P calc(i)" is calculated based on the following formula (4).

[0195] P calc (i)=P pred (i)+w2·P pred (i)·d(i)···(4) The weight "w2" shown in the above formula (4) is, for example, a value greater than or equal to 0 and less than or equal to 1. The weight "w2" may be arbitrarily set by the evaluator, or may be set in advance.

[0196] As described above, the correction unit 158 ​​calculates the goal achievement probability "P pred The correction unit 158 ​​may further weight "d(i)" with a weight "w2". Then, the correction unit 158 ​​calculates the goal achievement probability "P pred (i)" and the goal achievement probability "P pred (i)” and the sum of them is the target achievement probability “P calc By adjusting the weight "w2", the evaluator determines whether the distance "d(i)" is equal to the target achievement probability "P calc The degree of influence on (i) can be changed arbitrarily.

[0197] As another example, the correction unit 158 ​​calculates the weighted goal achievement probability "P calc (i)" may be calculated.

[0198] P calc (i)=w3·P pred (i)+w2·P pred (i)·d(i)···(5) The weight "w3" shown in the above formula (5) is, for example, a value greater than or equal to 0 and less than or equal to 1. The weight "w3" may be arbitrarily set by the evaluator, or may be set in advance.

[0199] As shown in the above formula (5), the correction unit 158 ​​calculates the goal achievement probability "P predThen, the correction unit 158 ​​weights the target achievement probability "P pred (i)” and the goal achievement probability “P pred (i)” and the sum of them is the target achievement probability “P calc (i)" is output.

[0200] The weight "w2" shown in the above formula (4) or (5) may be automatically adjusted. In this case, the correction unit 158 ​​changes the weight "w2" based on the number of actual data 123 used during learning. As an example, the correction unit 158 ​​reduces the weight "w2" as the number of actual data 123 increases. In other words, the correction unit 158 ​​increases the weight "w2" as the number of actual data 123 decreases.

[0201] Furthermore, the weight "w3" shown in the above formula (5) may be automatically adjusted. In this case, the correction unit 158 ​​changes the weight "w3" based on the number of performance data 123 used during learning. As an example, the correction unit 158 ​​increases the weight "w3" as the number of performance data 123 increases. In other words, the correction unit 158 ​​decreases the weight "w3" as the number of performance data 123 decreases.

[0202] Furthermore, the sum of the weights "w2" and "w3" shown in the above formula (5) may be constant. calc The scale of (i) can be made constant. Typically, the sum of the weights "w2" and "w3" is 1. In this case, the weight "w3" follows the following formula (6).

[0203] w3=1-w2 (6)

[0204] (I3. Integration Section 359) Next, the function of the integrating unit 359 shown in Fig. 16 will be described with reference to Fig. 19. Fig. 19 is a diagram conceptually showing the integrating process performed by the integrating unit 359.

[0205] The integration unit 359 integrates the output result RS3 by the correction unit 158 and the output result RS1B by the achievement probability calculation unit 354. As an example, the integration unit 359 multiplies the target achievement probability "P calc (i)" defined in the output result RS3 and the target achievement probability "P'' calc (i)" defined in the output result RS1B to calculate the integration result RS5.

[0206] <J. Flowchart related to prediction processing> Next, referring to FIGS. 20 and 21, the flow of the prediction process according to the second embodiment will be described. FIG. 20 is a flowchart showing the flow of the prediction process according to the second embodiment.

[0207] The flowchart shown in FIG. 20 is different from the flowchart shown in FIG. 12 in that it includes the process of step S200C. Since the other points are as described above, those descriptions will not be repeated.

[0208] FIG. 21 is a flowchart showing the flow of the process of step S200C shown in FIG. 20. Referring to FIG. 21, the process of step S200C will be described.

[0209] In step S230, the control device 101 initializes the variable "i". As an example, the variable "i" is initialized to "1".

[0210] In step S232, the control device 101 functions as the above-described distance weight calculation unit 156, and acquires the i-th input data 125 included in the prediction data set 124. Next, for each of the performance data 123 included in the learning data set 122, the control device 101 calculates the distance from the i-th input data 125. Since the method for calculating the distance is as described above, the description thereof will not be repeated.

[0211] In step S234, the control device 101 functions as the distance weight calculation unit 156 described above, and specifies the minimum distance "d0(i)" from the distances calculated in step S232.

[0212] In step S240, the control device 101 determines whether to end the calculation process of the minimum distance "d0(i)". The end condition of the calculation process is the same as that in step S220. When the control device 101 determines to end the calculation process of the minimum distance "d0(i)" (YES in step S240), it switches the process to step S250. Otherwise (NO in step S240), the control device 101 switches the control to step S242.

[0213] In step S242, the control device 101 increments the variable "i". That is, the control device 101 increases the variable "i" by 1. After that, the control device 101 returns the control to step S232.

[0214] In step S250, the control device 101 functions as the distance weight calculation unit 156 described above, normalizes the minimum distance "d0(i)" so that it is 0 or more and 1 or less, and calculates the normalized distance "d(i)". Since the normalization process is as described above, the description thereof will not be repeated.

[0215] In step S252, the control device 101 functions as the correction unit 158 described above, and weights the target achievement probability "P pred (i)" calculated in step S216 with a weight corresponding to the normalized distance "d(i)" calculated in step S250. Thereby, the control device 101 calculates the weighted target achievement probability "P calc (i)". Since the calculation process is as described above, the description thereof will not be repeated.

[0216] After that, the control device 101 exits the process of step S200C and switches the control to S330.

[0217] <K. Others> The above-described integrating unit 359 integrates the target achievement probabilities by equally multiplying the target achievement probabilities for each physical property. However, some physical properties to be evaluated may be more important than others. Therefore, the integrating unit 359 may weight the target achievement probabilities for each physical property before integrating the target achievement probabilities.

[0218] As an example, the achievement probability after integration is calculated based on the following formula (7).

[0219] log(P calc_all (i))=log(P calc_1 (i))·w1+log(P calc_2 (i))·w2+···+log(P calc_n (i))·w n ···(7) The log(P calc_all (i))" indicates the logarithm of the integrated probability of achieving the target for all properties after weighting. calc_n (i))" indicates the logarithm of the target achievement probability for the nth physical property. n " indicates the weight set for the nth physical property. "n" indicates a positive integer. "w n " may be arbitrarily set by the evaluator or may be set in advance. Typically, "w n The sum of " is 1.

[0220] The embodiments disclosed herein should be considered to be illustrative in all respects and not restrictive. The scope of the present invention is defined by the claims, not by the above description, and is intended to include all modifications within the meaning and scope of the claims. [Explanation of symbols]

[0221] 100 information processing device, 101 control device, 102 ROM, 103 RAM, 104 communication interface, 105 display interface, 106 display device, 107 input interface, 108 input device, 115 bus, 120 auxiliary storage device, 122 learning dataset, 123 actual data, 124 prediction dataset, 125 input data, 126 classification model, 127 regression model, 128 learning program, 130 prediction program, 132 goal condition, 152 learning unit, 154 achievement probability calculation unit, 156 distance weight calculation unit, 158 correction unit, 326 Gaussian process regression model, 328 learning program, 330 prediction program, 332 goal condition, 352 learning unit, 354 achievement probability calculation unit, 359 integration unit, 360 output unit, G1 graph, NR probability distribution, RS1 Output result, RS1A output result, RS1B output result, RS2 output result, RS3 output result, RS5 integrated result.

Claims

1. A method performed by an information processing device, comprising: inputting input data including raw material blending conditions into the trained classification model, and obtaining the probability that a material produced under the blending conditions will be classified into each of a plurality of evaluation results defined for a predetermined first evaluation test; a step of predicting a first target achievement probability that the material satisfies a desired first condition based on the plurality of probabilities obtained in the obtaining step; inputting the input data into a trained regression model and obtaining a probability distribution of predicted values ​​of physical properties of the material to be evaluated in a second evaluation test that is different from the first evaluation test; predicting a second target achievement probability that the material satisfies a desired second condition based on the probability distribution obtained in the obtaining step; and combining the first goal achievement probability and the second goal achievement probability by multiplying them together.

2. the classification model is generated by a learning process using a plurality of performance data; The method according to claim 1 , wherein each of the plurality of performance data associates a blending condition of raw materials with an evaluation result of the first evaluation test performed on a material produced under the blending condition.

3. the regression model is generated by a learning process using the plurality of performance data; 3. The method according to claim 2, wherein each of the plurality of performance data associates a blending condition of raw materials, an evaluation result of the first evaluation test conducted on a material produced under the blending condition, and an evaluation result of the second evaluation test conducted on a material produced under the blending condition.

4. the regression model is a Gaussian process regression model; The method according to any one of claims 1 to 3, wherein the Gaussian process regression model is trained to receive input of blending conditions of raw materials and output a probability distribution of predicted values ​​for the physical properties.

5. the first evaluation test is a qualitative evaluation test, The method according to any one of claims 1 to 3, wherein the second evaluation test is a quantitative evaluation test.

6. In the integrating step, the first target achievement probability and the second target achievement probability are multiplied for each of the plurality of input data for which different blending conditions are specified, The method according to any one of claims 1 to 3, further comprising the step of outputting, from among the plurality of input data, the input data that produces the greatest integration result in the integrating step.

7. The method according to any one of claims 1 to 3, wherein the blending conditions include at least one of a blending ratio of raw materials and a manufacturing condition when blending the raw materials.

8. The method of any one of claims 1 to 3, wherein the material comprises a tacky adhesive material.

9. The method further comprises: For each of the plurality of performance data, a step of calculating a similarity between a compounding condition defined in the performance data and a compounding condition defined in the input data; weighting the first goal achievement probability with weights according to the plurality of similarities calculated in the calculating step, The method according to claim 2 or 3, wherein the combining step multiplies the first goal achievement probability after weighting in the weighting step by the second goal achievement probability.

10. An information processing device, A control unit is provided, The control unit A process of inputting input data including raw material blending conditions into the trained classification model and obtaining the probability that a material produced under the blending conditions will be classified into each of a plurality of evaluation results specified for a predetermined first evaluation test; a process of predicting a first target achievement probability that the material satisfies a desired first condition based on the plurality of probabilities acquired in the process of acquiring; a process of inputting the input data into a trained regression model and obtaining a probability distribution of predicted values ​​of physical properties of an evaluation target in a second evaluation test, which is different from the first evaluation test, for the material; a process of predicting a second target achievement probability that the material satisfies a desired second condition based on the probability distribution acquired in the acquisition process; and integrating the first goal achievement probability and the second goal achievement probability by multiplying them together.

11. A program executed by an information processing device, The program is configured to: A process of inputting input data including raw material blending conditions into the trained classification model and obtaining the probability that a material produced under the blending conditions will be classified into each of a plurality of evaluation results specified for a predetermined first evaluation test; a process of predicting a first target achievement probability that the material satisfies a desired first condition based on the plurality of probabilities acquired in the process of acquiring; a process of inputting the input data into a trained regression model and obtaining a probability distribution of predicted values ​​of physical properties of an evaluation target in a second evaluation test, which is different from the first evaluation test, for the material; a process of predicting a second target achievement probability that the material satisfies a desired second condition based on the probability distribution acquired in the acquisition process; and integrating the first goal achievement probability and the second goal achievement probability by multiplying them together.

Images