Prediction system, prediction method, and program

Abstract

A prediction system, a prediction method, and a program for predicting the performance of a composition are provided. [Solution] A prediction system 100 includes a model generation unit that receives composition group data including a plurality of pairs of composition information, composition performance values, and performance classifications based on the performance of the compositions, and generates a plurality of prediction models that predict the performance values ​​from the composition information based on the performance classifications. The model generation unit includes a classification unit that classifies the composition group data based on the performance classifications, and a classification learning unit that generates a prediction model that predicts the performance value from the composition information for each of the composition group data classified based on the performance classifications.

Description

[Technical Field]

[0001] The present invention relates to a prediction system, a prediction method, and a program. [Background technology]

[0002] Patent Document 1 describes an apparatus comprising: "a composition acquisition unit that acquires composition data indicating the composition of a photosensitive resin composition; a characteristic acquisition unit that acquires characteristic data indicating the characteristics of the photosensitive resin composition; and a learning processing unit that executes a model learning process that uses the acquired composition data and learning data including the characteristic data to output recommended composition data indicating the composition of a photosensitive resin composition to be recommended in response to input of target characteristic data indicating the characteristics of a target photosensitive resin composition" (Claim 1). [Prior art document] [Patent documents] [Patent Document 1] JP 2020-77346 A Summary of the Invention

[0003] In a first aspect of the present invention, there is provided a prediction system for predicting the performance of a composition, the prediction system comprising: a model generation unit that receives as input composition group data including a plurality of pairs of composition information, composition performance values, and performance classifications that are classifications based on the performance of the composition, and generates a plurality of prediction models that predict the performance values ​​from the composition information based on the performance classifications.

[0004] The model generation unit may include a classification unit that classifies composition group data based on a performance classification, and a classification learning unit that generates a prediction model that predicts a performance value from composition information for each of the composition group data classified based on the performance classification.

[0005] The model generation unit may include a regression learning unit that generates a regression model that predicts a performance value from composition information based on the performance classification.

[0006] The model generation unit may include a regression coefficient acquisition unit that acquires regression coefficients in the regression model based on the performance classification.

[0007] The regression learning unit may generate a regression model including a probability distribution of the regression coefficients based on the performance classification, and the regression coefficient acquisition unit may acquire a representative value of the regression coefficients by sampling the regression model including the probability distribution of the regression coefficients.

[0008] The regression coefficient acquisition unit may perform sampling using the Markov Chain Monte Carlo method (MCMC).

[0009] The composition group data includes composition information and performance values ​​of the compositions, and may further include unclassified data that does not include a performance classification.

[0010] The regression learning unit may generate a regression model that predicts a performance value from composition information based on unclassified data by applying Gaussian mixture regression to the composition group data.

[0011] The performance value may be in the range of products producible with the composition.

[0012] The performance value may be the smallest dimension of a product that can be produced with the composition.

[0013] The performance classification may be a classification based on the appearance of the product produced from the composition.

[0014] The prediction system may further include a data generation unit that generates composition group data by assigning a performance classification to the composition based on an appearance image of a product manufactured using the composition.

[0015] The composition may be a resin composition, and the composition information may include information about polymers, monomers, and other components in the resin composition.

[0016] The prediction system may include a prediction unit that inputs composition information of a composition to be predicted into a prediction model and outputs a performance value based on the performance classification of the composition.

[0017] The prediction system may include a performance output unit that outputs the worst or best performance value among the performance values ​​based on the performance classification as the overall performance value of the composition.

[0018] In a second aspect of the present invention, there is provided a prediction system for predicting the performance of a composition, comprising a model generation unit that receives composition group data including a plurality of pairs of composition information and performance values ​​of the composition, and generates a plurality of prediction models that predict performance values ​​from the composition information.

[0019] In a third aspect of the present invention, there is provided a prediction system comprising: a prediction unit that inputs composition information of a composition to be predicted into a prediction model and outputs a performance value based on a performance classification, which is a classification based on the performance of the composition; and a performance output unit that outputs the performance value based on the performance classification that has the worst or best result as the overall performance value of the composition.

[0020] In a fourth aspect of the present invention, there is provided a prediction method, including: inputting composition group data including a plurality of pairs of composition information, a performance value of the composition, and a performance classification that is a classification based on the performance of the composition, to the prediction system of any of the above-mentioned first aspects; and executing a model generation step of generating a plurality of prediction models that predict the performance value from the composition information based on the performance classification.

[0021] In a fifth aspect of the present invention, the prediction system according to the second aspect includes a model generation step of inputting composition group data including a plurality of pairs of composition information and performance values ​​of the compositions and generating a plurality of prediction models for predicting performance values ​​from the composition information; The present invention provides a prediction method comprising:

[0022] In a sixth aspect of the present invention, there is provided a prediction method in which the prediction system of the third aspect described above executes a prediction step of inputting composition information of a composition to be predicted into a prediction model and outputting a performance value based on a performance classification, which is a classification based on the performance of the composition, and a performance output step of outputting the performance value based on the performance classification, which has the worst or best result, as an overall performance value of the composition.

[0023] In a seventh aspect, the present invention provides a program that, when executed by a computer, causes the computer to function as any one of the prediction systems described above.

[0024] The above summary of the invention does not list all of the features of the present invention, and subcombinations of these features may also be inventions. [Brief explanation of the drawings]

[0025] [Figure 1] 1 shows the configuration of a prediction system 100 according to this embodiment. [Figure 2] 1 shows an example of a model generation unit 130 according to this embodiment. [Figure 3] 1 shows an example of a manufacturing apparatus 200 according to this embodiment. [Figure 4] 1 shows a flow of a prediction method according to the present embodiment. [Figure 5] An example of a subflow of S10 in FIG. 4 is shown below. [Figure 6] An example of composition group data is shown below. [Figure 7] An example of a subflow of S20 in FIG. 4 is shown below. [Figure 8] An example of composition group data classified by performance category is shown below. [Figure 9] 5 shows another example of the subflow of S20 in FIG. [Figure 10A] An example of a regression model is shown below. [Figure 10B] An example of a representative value of the regression coefficient is shown below. [Figure 11]10 shows an example of a predicted performance value output by the prediction system 100. [Figure 12] 10 shows a modified example of the flow of the prediction method according to the present embodiment. [Figure 13] An example of the subflow of S10' in FIG. 12 is shown below. [Figure 14] An example of the subflow of S20' in FIG. 12 is shown below. [Figure 15] 10 shows a modified example of the flow of the prediction method of the present embodiment. [Figure 16] An example of the subflow of S20'' in FIG. 15 is shown below. [Figure 17] 22 illustrates an example computer 2200 in which aspects of the present invention may be embodied, in whole or in part. DETAILED DESCRIPTION OF THE INVENTION

[0026] The present invention will be described below through embodiments of the invention, but the following embodiments do not limit the scope of the invention according to the claims. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention.

[0027] 1 shows the configuration of a prediction system 100 according to this embodiment. The prediction system 100 generates a prediction model for predicting the performance of a composition. The prediction system 100 predicts the performance value of the composition to be predicted using the prediction model.

[0028] The composition is not particularly limited, and may be, for example, a resin composition or a photosensitive resin composition. The photosensitive resin composition may be a photocurable negative type or a photodissolvable positive type.

[0029] The photosensitive resin composition may be in a liquid state, or may be provisionally cured and / or dried. The liquid photosensitive resin composition may be a liquid resist in which a photosensitive resin is dissolved in a solvent. The provisionally cured and / or dried photosensitive resin composition may be in the form of a film formed on a support layer. In this case, it is also called a photosensitive resin laminate.

[0030] As an example in this embodiment, the photosensitive resin composition may be formed into a roll by being laminated with a support layer and a protective layer and wound up for the purpose of facilitating storage and transportation, and in this case, the roll-shaped laminate is also referred to as a photosensitive resin roll.

[0031] The prediction system 100 includes a database 110, a data generation unit 120, a model generation unit 130, a prediction unit 140, and a performance output unit 150. The prediction system 100 may include modules having other functions as necessary.

[0032] The prediction system 100 may be a computer such as a PC (personal computer), a tablet computer, a smartphone, a workstation, a server computer, or a general-purpose computer, or may be a computer system in which multiple computers are connected.

[0033] Alternatively, the prediction system 100 may be a dedicated computer designed for purposes such as generating a prediction model, or may be dedicated hardware implemented using dedicated circuits. The prediction system 100 may be implemented by a single device (computer), or may be implemented by multiple devices with different roles. Although not specifically described below, the prediction system 100 is equipped with a memory / hard disk, etc., which stores information necessary for processing as appropriate and transmits information between processing modules such as the data generation unit 120 and the model generation unit 130.

[0034] The database 110 may store data required for the prediction system 100 and provide the data to each module of the prediction system 100. For example, the database 110 may store composition information of compositions, performance values ​​of compositions, information on the performance of compositions for classifying the performance of compositions, and prediction models. The database 110 may be constructed in a storage device such as a memory or a hard disk.

[0035] The data generating unit 120 generates composition group data for the compositions. The composition group data includes a plurality of pairs of composition information and performance values ​​of the compositions. The composition group data may include a plurality of pairs of composition information, performance values ​​of the compositions, and performance classifications that are classifications based on the performance of the compositions. The composition group data may include composition information and performance values ​​of the compositions, but may include unclassified data that does not include performance classifications.

[0036] The data generating unit 120 may generate the composition group data by assigning a performance classification to the composition based on the performance of the composition itself, or may generate the composition group data by assigning a performance classification to the composition based on the performance of a product manufactured from the composition. The performance of the product may be performance based on its appearance or may be based on an appearance image.

[0037] The model generation unit 130 acquires composition group data and generates a prediction model that predicts a performance value from the composition information. The prediction model may be a regression model that predicts a performance value from the composition information. The model generation unit 130 may generate multiple prediction models based on performance classifications. The model generation unit 130 may generate a prediction model for each performance classification.

[0038] The prediction unit 140 inputs composition information of the composition to be predicted into a prediction model and outputs a performance value of the composition. The prediction unit 140 may output a performance value for each of multiple prediction models. The prediction unit 140 may output a performance value based on a performance classification of the composition. The prediction model may be a prediction model generated by the model generation unit 130.

[0039] The performance output unit 150 outputs an overall performance value of the composition based on the performance values ​​of the composition output by the prediction unit 140. The overall performance value may be a performance value selected from the performance values ​​based on the performance classification. For example, the performance output unit 150 may output the performance value based on the performance classification that has the worst or best result as the overall performance value of the composition.

[0040] 2 shows an example of the model generation unit 130 according to this embodiment. The model generation unit 130 may include a classification unit 132, a classification learning unit 134, a regression learning unit 136, and a regression coefficient acquisition unit 138.

[0041] The classification unit 132 classifies the composition group data based on performance classification. The performance classification may be determined based on information on the performance of the composition (sometimes simply referred to as "performance information"), including at least one of the performance and / or physical properties of the composition itself, the performance and / or physical properties of the composition after processing the composition in a predetermined manner, or the performance and / or physical properties of a product manufactured from the composition. For example, the performance classification may be a classification based on the appearance of a product manufactured from the composition.

[0042] The classification learning unit 134 generates a prediction model that predicts a performance value from the composition information for each of the composition group data classified based on the performance classification. The classification learning unit 134 may generate the prediction model using a regression model, a neural network model such as a recurrent or time-delay type, a random forest model, a gradient boosting model, a logistic regression model, or a support vector machine (SVM) model.

[0043] The regression learning unit 136 generates a regression model that predicts a performance value from the composition information. The regression model may be a linear regression model such as simple regression, multiple regression, or logistic regression. The regression model may also be a nonlinear regression model. The regression learning unit 136 may generate a regression model that includes a probability distribution of regression coefficients.

[0044] The regression learning unit 136 may generate a regression model that predicts a performance value from composition information by applying Gaussian mixture regression to the composition group data. The regression learning unit 136 may apply Gaussian mixture regression to composition group data that includes unclassified data. The regression learning unit 136 may also apply Gaussian mixture regression to composition group data to which performance classifications have been assigned.

[0045] The regression coefficient acquisition unit 138 acquires regression coefficients in the regression model. The regression coefficients may be different for each regression model. The regression coefficient acquisition unit 138 may acquire regression coefficients for each performance classification. The regression coefficient acquisition unit 138 may acquire representative values ​​of the regression coefficients by sampling a regression model including a probability distribution of the regression coefficients. The regression coefficient acquisition unit 138 may perform sampling using a Markov Chain Monte Carlo method (MCMC).

[0046] The model generating unit 130 may not include any of the classifying unit 132, the classification learning unit 134, the regression learning unit 136, and the regression coefficient acquiring unit 138 described above.

[0047] 3 shows an example of a manufacturing apparatus 200 according to this embodiment. The manufacturing apparatus 200 manufactures a composition. The manufacturing apparatus 200 may include a mixing section 210 that mixes raw materials for the composition, and a production section 220 that produces the composition from the mixed raw materials.

[0048] The mixing unit 210 may perform mixing based on composition information of the composition whose performance value has been predicted. For example, the mixing unit 210 may mix raw materials based on the composition information of a composition whose performance value and / or overall performance value falls within a predetermined range, among compositions whose performance values ​​have been predicted from the composition information by the prediction system 100.

[0049] The generating section 220 may include a filtering section that filters the raw materials, an application section that applies the raw materials onto a support layer to form a film of the photosensitive resin composition, and a roller section that attaches a protective layer onto the photosensitive resin composition and winds up the laminate.

[0050] According to this embodiment, it is possible to predict the performance value of a composition with high accuracy using multiple prediction models from composition information of the composition. This makes it possible to design a composition having a target performance value in a short period of time. Furthermore, by using a prediction model based on performance classification based on the performance of the composition as the prediction model, it is possible to predict the composition information of a composition that will achieve a target performance value with high accuracy in a short period of time. Furthermore, according to this embodiment, it is possible to reduce the computational resources and storage resources used to predict a composition that will achieve a target performance value.

[0051] 4 shows a flow of the prediction method according to this embodiment. The prediction system 100 predicts a performance value from composition information of a composition by executing each process from S10 to S50. The order of S10 to S50 may be changed, and some processes may be omitted.

[0052] In S10, the prediction system 100 acquires composition group data for the composition. The composition group data includes a plurality of pairs of composition information and performance values ​​of the composition. In addition to the above pairs, the composition group data further includes performance information for classifying the performance of the composition.

[0053] Fig. 5 shows an example of a subflow of S10 in Fig. 4. The prediction system 100 may perform S10 by executing the flow of S120 to S140 in Fig. 5.

[0054] First, in S120, the data generating unit 120 acquires composition information, a performance value, and performance information of the composition. The data generating unit 120 may acquire the composition information, the performance value, and the performance information from the database 110.

[0055] The composition information is information about the components contained in the composition and / or the raw materials that can produce the composition (hereinafter, the components and raw materials are collectively referred to as "components, etc.").

[0056] The composition information may be, for example, the presence or absence of a component, etc. in the composition, the content of the component, etc., and / or the content ratio of the component, etc. The component, etc. may be represented by information that can at least partially identify its chemical structure, and may be represented by, for example, a common name, a common name, an IUPAC name, a trade name, a model number, a chemical formula, a structural formula, a composition formula, a CAS number, a three-dimensional structure, or a molecular fingerprint.

[0057] The composition may be, for example, a resin composition, for example, a photosensitive resin composition. However, the composition is not limited thereto. The composition information may include, as components, information on polymers, monomers, and other components in the resin composition.

[0058] More specifically, the components may include at least one of an alkali-soluble polymer, an ethylenically unsaturated bond-containing compound, a photopolymerization initiator, a resin having a repeating unit containing an acid-decomposable group (for example, a group that is deprotected by an acid), a phenolic resin, a photoacid generator, a dissolution inhibitor, a sensitizer, a polymerization inhibitor, an adhesion agent, a plasticizer, and a solvent.

[0059] Among these, the alkali-soluble polymer may be a polymer in a photocurable resin composition, such as a polymer containing a carboxyl group. The ethylenically unsaturated bond-containing compound may be a monomer in a photocurable resin composition. The photopolymerization initiator may bond monomers together in a photocurable resin composition upon exposure to light. Resins having repeating units containing acid-decomposable groups and phenolic resins may be polymers in a photosoluble resin composition and may be dissolved and decomposed by acid. The photoacid generator generates acid in a photosoluble resin composition upon exposure to light. The dissolution inhibitor, also known as a dissolution inhibitor, inhibits the dissolution of components in an alkaline aqueous solution. The sensitizer may be, for example, a photosensitizer, but other types of sensitizers such as N-phenylglycine may also be used. The polymerization inhibitor may inhibit polymerization reactions due to the influence of light or heat. The adhesion agent enhances the adhesion of the photosensitive resin composition to the substrate surface. The plasticizer is added to the photosensitive resin composition to impart flexibility and facilitate processing.

[0060] The performance value may be a numerical representation of the performance and / or physical properties of the composition itself, the performance and / or physical properties of the composition after processing by a predetermined method, or the performance and / or physical properties of a product produced from the composition. For example, the performance value may be at least one of the minimum development time of the composition as a photosensitive resin (hereinafter also referred to as a "photosensitive resin composition"), sensitivity to light, transparency, resolution, the range of products that can be produced from the composition, adhesion to a substrate, foaming property of a developer, cohesive property of a developer, edge fuse characteristics, flexibility of a cured film, color stability, peel time, size of a peeled piece, film thickness, tackiness with a support layer or a protective layer, and tentability.

[0061] Development of a photosensitive resin composition may involve exposing the photosensitive resin composition to light to cure or solubilize the photosensitive resin composition in the exposed regions, and then removing the photosensitive resin composition in the exposed or unexposed regions to reveal a negative or positive image corresponding to the exposed region. When the photosensitive resin composition is photocurable (also referred to as negative), the minimum development time refers to the minimum time required for the photosensitive resin composition in the unexposed regions to be developed. When the photosensitive resin composition is photosoluble (also referred to as positive), the minimum development time refers to the minimum time required for the photosensitive resin composition in the exposed regions to be developed. For example, when the photosensitive resin composition contains an alkali-soluble polymer, the minimum development time may be the time required for all of the photosensitive resin composition to be removed by developing the photosensitive resin composition coated or laminated on a substrate by spraying an alkaline solution without exposure. The minimum development time may be the time required when parameters that may affect the development time, such as the type of developer, the concentration of the developer, the number of spray nozzles, and the spray pressure, are fixed.

[0062] When the photosensitive resin composition is photocurable, the sensitivity to light is also referred to as the minimum curing exposure amount, and indicates the minimum exposure amount capable of forming an image corresponding to the exposed area. For example, when the photosensitive resin composition is photocurable, the sensitivity to light is the minimum exposure amount (mJ / cm) at which the photosensitive resin composition that is applied or laminated to a substrate is cured and remains on the substrate when exposed and developed.2 As an example, the sensitivity to light may be calculated from the lowest transmittance at which the photosensitive resin composition is cured when the photosensitive resin composition is exposed to light using a mask with gradually varying transmittance.

[0063] The transmittance, also referred to as the transmittance, indicates the transmittance of light. The wavelength of the transmitted light may be a wavelength that cures or solubilizes the photosensitive resin composition.

[0064] The range of products producible from the composition may be the dimensions that can be formed without defects using the photosensitive resin composition, etc. For example, the range of products producible from the composition may be the minimum dimensions of products producible from the composition.

[0065] The developer foaming property indicates the foaming property when a photosensitive resin composition is developed using a developer. As the developer foaming property, values ​​measured by various known methods may be used.

[0066] The developer coagulation property indicates the coagulation property when the photosensitive resin composition is developed using a developer. As the developer coagulation property, a value measured by various known methods may be used.

[0067] The edge fuse characteristic indicates the amount of photosensitive resin composition that protrudes from the end face of the photosensitive resin roll due to winding pressure during storage of the photosensitive resin roll. The smaller the edge fuse characteristic, the longer the usable life of the photosensitive resin laminate, which is preferable. Values ​​measured by various known methods may be used as the edge fuse characteristic.

[0068] The flexibility of the cured film indicates the flexibility of the developed photocurable photosensitive resin composition. For example, the flexibility of the cured film may be the smallest diameter at which the resist does not crack when a photosensitive resin laminate of the photosensitive resin composition laminated on a flexible substrate is exposed to light and developed, and then wrapped around multiple cylinders of different diameters. As an example, the flexibility of the cured film may be measured using a mandrel bending tester.

[0069] The tackiness with the support layer or protective layer indicates the adhesiveness between the photosensitive resin laminate and the support layer or protective layer. For example, the tackiness may be the force required to peel the photosensitive resin laminate from the support layer or protective layer using a Tensilon device. If the tackiness is too high or too low, the usability of the photosensitive resin wound body will be reduced, so it is preferable that the tackiness be within an appropriate range.

[0070] The peeling time indicates the releasability when peeling the photosensitive resin composition from the substrate. For example, the peeling time may be the time it takes for the photosensitive resin composition to peel from the substrate when a photosensitive resin laminate laminated on a substrate and exposed to light is immersed in an alkaline stripping solution for development. A shorter peeling time is preferred.

[0071] The color stability indicates the stability of the color tone of the photosensitive resin composition. As the color stability, values ​​measured by various known methods may be used.

[0072] The peeled piece size indicates the size of the photosensitive resin composition (also referred to as peeled piece) peeled from the substrate. The peeled piece size may be the size of the peeled piece peeled from the substrate and broken into small pieces by a water spray. Smaller peeled pieces are preferred.

[0073] The tentability indicates the rate of breakage when tenting tent holes in a substrate with a photosensitive resin laminate. For example, the tentability may be the number of broken tent holes when a photosensitive resin laminate is laminated onto a substrate having tent holes with a diameter of 1 to 10 mm, exposed to light, and developed. The smaller the tentability, the better.

[0074] The appearance image may be an image of the composition or an image of the product. The appearance image may be an image of the appearance of a component formed from the composition in the product. The appearance image may be an image of the surface of the product or a micrograph of the surface of the product.

[0075] Next, in S140, the data generation unit 120 assigns a performance classification to the composition based on the performance information of the composition. The performance information may be, for example, physical properties of the composition, such as the viscosity of a solution containing the composition, and the appearance of a product manufactured from the composition. The data generation unit 120 may acquire an appearance image of the product manufactured from the composition, and assign a performance classification to a pair of composition information and performance value of the composition based on the appearance image of the product manufactured from the composition.

[0076] The data generating section 120 may assign a performance classification to a plurality of pairs of composition information and performance values ​​of the composition, where the composition information and performance values ​​are similar to each other, using a common appearance image.

[0077] The data generating unit 120 may extract from the appearance image appearance characteristics such as shape, structure, and color formed on the product depending on the components of the composition, etc., and assign a performance classification based on the above characteristics. The data generating unit 120 may assign the same performance classification to compositions that have the same appearance characteristics (for example, the same uneven shape on the resin surface) and compositions that have the same color.

[0078] The appearance characteristic of the product used for performance classification may be a characteristic preselected from characteristics appearing in the appearance of the product. The appearance characteristic of the product may be a characteristic correlated with the performance value of the composition. For example, if the performance value of the composition is developer foaming property, the appearance characteristic may be a characteristic that appears in the product in response to an increase in an evaluation index of developer foaming property (e.g., foam height after dissolution in the developer and stirring). If the performance value of the composition is peelability, the appearance characteristic may be a characteristic that appears in the product in response to a delay in peeling time. The appearance characteristic used for performance classification may be, for example, a characteristic that has a certain level of correlation with the performance value, as determined by performing correlation analysis, regression analysis, etc. on the characteristic and the performance value of the composition, and then used for performance classification.

[0079] In the flow of S10 in Fig. 4, a performance classification may be assigned in advance to at least some pairs of composition information and performance values. In this case, the above-described flow of S120 to S140 may be omitted for pairs to which a performance classification has been assigned in advance.

[0080] Fig. 6 shows an example of composition group data 300. In the example shown in Fig. 6, the composition group data 300 includes pairs of composition information 310, performance value 320, and performance classification 330 of compositions (No. 1 to 5).

[0081] The composition information 310 may include information on the polymers (polymer 1 and polymer 2) and monomers (monomer 1) in the composition (resin composition). Polymer 1 and polymer 2 represent specific types of polymers. Monomer 1 represents a specific type of monomer. The performance classification 330 represents the appearance characteristics (A, B, and C in the example of FIG. 6) of the product produced from the composition and the performance classification 330 assigned to the appearance characteristics (No. 1 to 3 in the example of FIG. 6).

[0082] In the example shown in Figure 6, a composition that produces a product exhibiting characteristic A in appearance is assigned performance classification No. 1, a composition that produces a product exhibiting characteristic A+B (having both characteristics A and B) is assigned performance classification No. 2, and a composition that produces a product exhibiting characteristic C is assigned performance classification No. 3.

[0083] In Figure 6, for example, composition No. 1 contains 0.5 parts by weight of a polymer of the type "Polymer 1," 0.1 parts by weight of a polymer of the type "Polymer 2," and 0.4 parts by weight of a monomer of the type "Monomer 1," resulting in a performance value of 4.2. Furthermore, composition No. 1 is assigned performance classifications No. 1 and No. 2 in accordance with the fact that products made from that composition exhibit appearance characteristics A and B.

[0084] After the process of S10, in S20, the prediction system 100 generates a prediction model that predicts a performance value from composition information based on the performance classification.

[0085] Fig. 7 shows an example of a subflow of S20 in Fig. 4. The model generation unit 130 may perform S20 by executing the flow from S220 to S240 in Fig. 7.

[0086] First, in S220, the classification unit 132 classifies the composition group data based on the performance classification. The classification unit 132 may classify the composition group data for each performance classification. The classification unit 132 may classify the composition group data by grouping multiple performance classifications into one group.

[0087] Fig. 8 shows an example of composition group data 400 after classification by performance category. In the example shown in Fig. 8, the composition group data 300 shown in Fig. 6 is classified by performance category. The classified composition group data 400 includes composition information 410 and performance values ​​420 classified by performance category 430. The composition information 410 includes information on the polymers (polymer 1 and polymer 2) and monomers (monomer 1) in the composition (resin composition).

[0088] In the example shown in Figure 8, for example, compositions Nos. 1, 2, 3, and 5 are classified into performance category 430 No. 1 (including appearance feature A of the product), compositions Nos. 1 and 3 are classified into performance category 430 No. 2 (including appearance features A and B), and compositions Nos. 2 to 4 are classified into performance category 430 No. 3 (including appearance feature C).

[0089] Next, in S240, the classification learning unit 134 generates a prediction model that predicts a performance value from the composition information for each of the composition group data classified based on the performance classification. For example, in the example of FIG. 8, a prediction model may be generated using compositions Nos. 1, 2, 3, and 5 classified into performance classification No. 1 430, and another prediction model may be generated using compositions Nos. 2 to 4 classified into composition classification No. 3 430. The classification learning unit 134 may generate the prediction model using a regression model, a neural network model such as a recurrent or time-delay type, a random forest model, a gradient boosting model, a logistic regression model, or a support vector machine (SVM) model.

[0090] When classification learning unit 134 generates a prediction model using a regression model, classification learning unit 134 may function as regression learning unit 136, which will be described later.

[0091] Fig. 9 shows another example of the subflow of S20 in Fig. 4. The model generation unit 130 may perform S20 by executing the flow of S220 and S260 to S280 in Fig. 9.

[0092] First, in S220, the classification unit 132 classifies the composition group data based on the performance classification. The description of S220 in FIG. 7 may be applied as is to the description of S220.

[0093] Next, in S260, the regression learning unit 136 generates a regression model that predicts a performance value from the composition information based on the performance classification. The regression model may be a linear regression model such as simple regression, multiple regression, or logistic regression. The regression model may also be a nonlinear regression model. The regression learning unit 136 may generate a regression model including a probability distribution of regression coefficients based on the performance classification.

[0094] In S260, for example, for the composition group data in the example shown in FIG. 8, a regression model may be generated for each performance classification using composition information classified into each of performance classifications No. 1 to 3 as explanatory variables and performance values ​​as response variables.

[0095] Next, in S280, the regression coefficient acquisition unit 138 acquires a representative value of the regression coefficients in the regression model. The regression coefficient acquisition unit 138 may acquire the representative value of the regression coefficients by sampling a regression model including a probability distribution of the regression coefficients. The regression coefficient acquisition unit 138 may perform sampling using a Markov Chain Monte Carlo method (MCMC). The representative value may be the mean, median, or expected value of the sampled posterior probability distribution. If a prediction model with high prediction accuracy is acquired in S260, S280 may be omitted.

[0096] FIG. 10A shows an example of a regression model. In FIG. 10A, graph 500 represents an image in which composition group data (composition information and performance values) of a composition are plotted on a coordinate system. In graph 500, the horizontal axis (x-axis) represents the ratio of polymer 1 to the total amount of polymer 1 and polymer 2 contained in the composition, and the vertical axis (z-axis) represents the performance value. Although not shown in the figure, the y-axis (vertical direction on the paper) represents the ratio of the monomer contained in the composition. In S260, the regression learning unit 136 generates a regression model for each performance classification using composition group data 510 and composition group data 520 classified by performance classification. At this time, the regression learning unit 136 may generate regression models 530 and 540 including the probability distribution of the regression coefficients (the area surrounded by the dotted line in FIG. 10A).

[0097] 10B shows an example of the representative values ​​of the regression coefficients. The regression coefficient acquisition unit 138 may perform sampling on the regression model 530 and the regression model 540 including the probability distribution of the regression coefficients shown in FIG. 10A to acquire the representative value 550 of the regression coefficients in the regression model 530 and the representative value 560 of the regression coefficients in the regression model 540. The model generation unit 130 may predict the performance value using the regression model 570 including the representative value 550 of the regression coefficients and the regression model 580 including the representative value 560 of the regression coefficients (both shown by solid lines in FIG. 10A).

[0098] After the process of S20, in S30, the prediction unit 140 inputs composition information of the composition to be predicted into the prediction model. If the model generation unit 130 has generated multiple prediction models, the composition information may be input into one or more of the multiple prediction models.

[0099] Next, in S40, the prediction unit 140 outputs a performance value based on the performance classification predicted by the prediction model (sometimes referred to as a "predicted performance value"). The prediction unit 140 may output a predicted performance value for each performance classification.

[0100] Next, in S50, the performance output unit 150 outputs an overall predicted performance value of the composition. If performance values ​​based on multiple prediction models are output in S40, the performance output unit 150 may output a performance value selected from the performance values ​​for each of the multiple prediction models as the overall predicted performance value in S50. For example, the predicted performance value for each performance category that yields the worst or best result may be used as the overall predicted performance value for the composition.

[0101] 11 shows an example of predicted performance values ​​output by the prediction system 100. The example shown in FIG. 11 shows composition information 610 to be predicted, predicted performance values ​​620 for each performance class, and an overall predicted performance value 630. In the example of FIG. 11, for composition information 610 of composition No. 1, a predicted performance value of 7.3 for performance class A, a predicted performance value of 5.1 for performance class B, and a predicted performance value of 9.3 for performance class C are obtained, and the predicted performance value of performance class C, which has the largest numerical value (e.g., the worst result) among the predicted performance values ​​of performance classes A to C, may be output as the overall predicted performance value 630.

[0102] If the predicted performance value output in S40 or the overall predicted performance value output in S50 is not within a predetermined range, the prediction system 100 may execute the steps S30, S40, and S50 again using different composition information as the composition information to be predicted, and may repeat the steps S30 to S50 until the target predicted performance value is output.

[0103] According to this embodiment, the performance value of a composition can be predicted with high accuracy from composition information of the composition using multiple prediction models based on performance classification based on the performance of the composition. This makes it possible to design a composition having a target performance value in a short period of time. Furthermore, by using a prediction model based on performance classification based on the performance of the composition as the prediction model, it is possible to predict composition information of a composition that will achieve a target performance value with high accuracy in a short period of time. Furthermore, according to this embodiment, it is possible to reduce the computational resources and storage resources used to predict a composition that will achieve a target performance value.

[0104] Next, a modification of this embodiment will be described.

[0105] Fig. 12 shows a modified example of the flow of the prediction method of this embodiment. The prediction system 100 predicts a performance value from composition information of a composition by executing each process from S10' to S50' in Fig. 12. The order of S10' to S50' may be changed, and some processes may be omitted.

[0106] First, in S10', the prediction system 100 acquires composition group data for the composition. The composition group data includes a plurality of pairs of composition information, performance values, and performance classifications for the composition, and further includes pairs of composition information and performance values ​​for compositions that do not include a performance classification as unclassified data. The prediction system 100 may classify compositions for which no external characteristics of the product were detected as unclassified data. The composition, composition information, performance values, product, performance classification, and external appearance image are substantially the same as those described with reference to FIG. 5, and therefore will not be described here.

[0107] Fig. 13 shows an example of a subflow of S10' in Fig. 12. The prediction system 100 may perform S10' by executing the flow of S120' to S140' in Fig. 13.

[0108] First, in S120', the data generation unit 120 acquires a pair of composition information and performance value of the composition and performance information of the composition for a portion of the composition (for example, a composition for which the appearance characteristics of a product manufactured from the composition have been detected). The flow of S120' is the same as S120 shown in Fig. 5, except that performance information is acquired for a portion of the composition, and therefore a description thereof will be omitted.

[0109] Next, in S140', the data generation unit 120 assigns a performance classification to the composition for which performance information has been acquired. The flow of S140' is the same as S140 shown in Fig. 5 except that a performance classification is assigned to part of the composition, and therefore a description thereof will be omitted.

[0110] After the process of S10', in S20', the prediction system 100 generates a prediction model that predicts a performance value from composition information based on performance classification and unclassified data.

[0111] Fig. 14 shows an example of a subflow of S20' in Fig. 12. The model generation unit 130 may perform S20' by executing the flow from S220' to S280' in Fig. 14.

[0112] First, in S220', the classification unit 132 classifies the composition group data for the compositions to which performance classifications have been assigned, based on the performance classifications. The flow of S220' is the same as S220 shown in Fig. 9, except that the composition group data is classified for the compositions to which performance classifications have been assigned, and therefore a description thereof will be omitted.

[0113] Next, in S240', the regression learning unit 136 generates a regression model for predicting a performance value from the composition information based on the performance classification for the composition group data to which the performance classification has been assigned. The flow of S240' is the same as S240 shown in Figure 9 except that a regression model is generated for the composition to which the performance classification has been assigned, and therefore a description thereof will be omitted.

[0114] Next, in S260', the regression learning unit 136 applies Gaussian mixture regression to the unclassified composition group data to generate a plurality of regression models that predict performance values ​​from the composition information of the unclassified data. The composition group data to which Gaussian mixture regression is applied in S260' may be at least a portion of the remaining composition group data used to generate the regression models in S240'. Even if performance classifications cannot be assigned to all compositions included in the composition group data, a regression model can be generated by applying Gaussian mixture regression to composition group data to which no performance classifications are assigned.

[0115] Next, in S280', the regression coefficient acquisition unit 138 acquires representative values ​​of the regression coefficients in the regression models generated in S240' and S260'. The flow of S280' is the same as S280 shown in Fig. 9 except that representative values ​​of the regression coefficients are acquired for regression models based on unclassified data in addition to regression models based on performance classification, and therefore a description thereof will be omitted.

[0116] After the process of S20', in S30', the prediction unit 140 inputs the composition information of the composition to be predicted into the prediction model. If the model generation unit 130 has generated multiple prediction models, the composition information may be input into one or more of the multiple prediction models.

[0117] Next, in S40', the prediction unit 140 outputs the predicted performance value predicted by the prediction model. The prediction unit 140 may output the predicted performance value for each regression model.

[0118] Next, in S50', the performance output unit 150 outputs an overall predicted performance value of the composition. If performance values ​​based on multiple prediction models are output in S40', the performance output unit 150 may output a performance value selected from the performance values ​​for each of the multiple prediction models as the overall predicted performance value in S50'. For example, the predicted performance value for each of the multiple prediction models that gives the worst or best result may be used as the overall predicted performance value of the composition.

[0119] The flow of S30', S40' and S50' is the same as S30, S40 and S50 shown in FIG. 4, except that the prediction model is changed to include a regression model based on unclassified data, so a description thereof will be omitted.

[0120] According to this embodiment, the performance value of a composition can be predicted with high accuracy from composition information using multiple prediction models based on performance classifications based on the performance of the composition and a regression model based on Gaussian mixture regression. This makes it possible to design a composition having a target performance value in a short period of time. Furthermore, by using a prediction model based on unclassified data that does not include performance classifications in addition to performance classifications based on the performance of the composition as a prediction model, it is possible to predict composition information of a composition that will achieve a target performance value from a wide range of compositions with high accuracy and in a short period of time. Furthermore, according to this embodiment, it is possible to reduce the computational resources and storage resources used to predict a composition that will achieve a target performance value.

[0121] Fig. 15 shows a modified example of the flow of the prediction method of this embodiment. The prediction system 100 predicts a performance value from composition information of a composition by executing each process from S10'' to S50'' in Fig. 15. The order of S10'' to S50'' may be changed, and some processes may be omitted.

[0122] First, in S10'', the prediction system 100 acquires composition group data for the composition. The composition group data includes multiple pairs of composition information and performance values ​​for the composition. Note that the composition group data in this modified example is unclassified data that does not include performance classification. The composition, composition information, and performance values ​​are substantially the same as those described with reference to FIG. 5, and therefore will not be described here.

[0123] After S10'', in S20'', the prediction system 100 generates a prediction model that predicts a performance value from the composition information.

[0124] Fig. 16 shows an example of a sub-flow of S20'' in Fig. 15. The model generation unit 130 may perform S20 by executing the flow from S220'' to S240'' in Fig. 16.

[0125] First, in S220", the regression learning unit 136 applies Gaussian mixture regression to the composition group data to generate a plurality of regression models that predict performance values ​​from composition information. In S220", the composition group data to which Gaussian mixture regression is applied includes unclassified data in which no performance classification is assigned to the composition information and performance values ​​of the compositions.

[0126] The regression model may be a linear regression model such as simple regression, multiple regression, logistic regression, etc. The regression model may also be a non-linear regression model. Each of the multiple regression models may include a probability distribution of the regression coefficients.

[0127] In S220'', the composition group data to which the mixed regression is applied may include data including performance classification. The regression learning unit 136 may perform Gaussian mixed regression using parameters set as hyperparameters based on the composition group data including the performance classification. Cross-validation may be performed on the regression model generated by the Gaussian mixed regression using the composition group data including the performance classification. This can improve the accuracy of prediction of the prediction model.

[0128] Next, in S240'', the regression coefficient acquisition unit 138 acquires a representative value of the regression coefficients in the regression model. The regression coefficient acquisition unit 138 may acquire the representative value of the regression coefficients by sampling a regression model including a probability distribution of the regression coefficients. The regression coefficient acquisition unit 138 may perform sampling using the Markov Chain Monte Carlo method (MCMC). The representative value may be the mean, median, or expected value of the sampled posterior probability distribution. If a prediction model with high prediction accuracy is acquired in S220'', S240'' may be omitted.

[0129] After processing S20", in S30", the prediction unit 140 inputs composition information of the composition to be predicted into the prediction model. If the model generation unit 130 has generated multiple prediction models, the composition information may be input into one or more of the multiple prediction models.

[0130] Next, in S40'', the prediction unit 140 outputs the predicted performance value predicted by the prediction model. The prediction unit 140 may output the predicted performance value for each regression model.

[0131] Next, in S50'', the performance output unit 150 outputs an overall predicted performance value of the composition. If performance values ​​based on multiple prediction models are output in S40'', the performance output unit 150 may output a performance value selected from the performance values ​​for each of the multiple prediction models as the overall predicted performance value in S50''. For example, the predicted performance value for each regression model that gives the worst or best result may be used as the overall predicted performance value of the composition.

[0132] The flow of S30'', S40'', and S50'' is the same as S30, S40, and S50 shown in FIG. 4 except that the prediction model is changed to a regression model based on unclassified data, so a description thereof will be omitted.

[0133] According to this embodiment, the performance value of a composition can be predicted with high accuracy from composition information of the composition using a regression model based on Gaussian mixture regression. This makes it possible to design a composition having a target performance value in a short period of time. Furthermore, by using a regression model based on Gaussian mixture regression as a prediction model, it is possible to predict composition information of a composition that will achieve a target performance value with high accuracy in a short period of time. Furthermore, according to this embodiment, it is possible to reduce the computational resources and storage resources used to predict a composition that will achieve a target performance value.

[0134] Various embodiments of the present invention may be described with reference to flowcharts and block diagrams, where the blocks may represent (1) stages of a process in which operations are performed or (2) sections of an apparatus responsible for performing the operations. Particular stages and sections may be implemented by dedicated circuitry, programmable circuitry provided with computer-readable instructions stored on a computer-readable medium, and / or a processor provided with computer-readable instructions stored on a computer-readable medium. Dedicated circuitry may include digital and / or analog hardware circuitry, and may include integrated circuits (ICs) and / or discrete circuits. Programmable circuitry may include reconfigurable hardware circuitry, including logical AND, OR, XOR, NAND, NOR, and other logical operations, flip-flops, registers, memory elements such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), and the like.

[0135] A computer-readable medium may include any tangible device capable of storing instructions that are executed by an appropriate device, such that the computer-readable medium having instructions stored thereon comprises an article of manufacture containing instructions that can be executed to create means for performing the operations specified in the flowcharts or block diagrams. Examples of computer-readable media may include electronic, magnetic, optical, electromagnetic, and semiconductor storage media. More specific examples of computer-readable media may include floppy disks, diskettes, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), electrically erasable programmable read-only memory (EEPROM), static random access memory (SRAM), compact disc read-only memory (CD-ROM), digital versatile disc (DVD), Blu-ray disc, memory stick, integrated circuit card, and the like.

[0136] The computer readable instructions may include either assembler instructions, Instruction Set Architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state-setting data, or source or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk®, JAVA®, C++, etc., and conventional procedural programming languages ​​such as the “C” programming language or similar programming languages.

[0137] The computer-readable instructions may be provided to a processor or programmable circuitry of a programmable data processing apparatus, such as a general-purpose computer, special-purpose computer, or other computer, either locally or over a wide-area network (WAN) such as a local area network (LAN), the Internet, etc., which executes the computer-readable instructions to create means for performing the operations specified in the flowcharts or block diagrams. Examples of processors include computer processors, processing units, microprocessors, digital signal processors, controllers, microcontrollers, etc.

[0138] 17 illustrates an example of a computer 2200 in which aspects of the present invention may be embodied, in whole or in part. Programs installed on the computer 2200 may cause the computer 2200 to function as or perform operations associated with an apparatus or one or more sections of the apparatus according to embodiments of the present invention, and / or to perform a process or steps of a process according to embodiments of the present invention. Such programs may be executed by the CPU 2212 to cause the computer 2200 to perform specific operations associated with some or all of the blocks in the flowcharts and block diagrams described herein.

[0139] A computer 2200 according to this embodiment includes a CPU 2212, a RAM 2214, a graphics controller 2216, and a display device 2218, which are interconnected by a host controller 2210. The computer 2200 also includes input / output units such as a communication interface 2222, a hard disk drive 2224, a DVD-ROM drive 2226, and an IC card drive, which are connected to the host controller 2210 via an input / output controller 2220. The computer also includes legacy input / output units such as a ROM 2230 and a keyboard 2242, which are connected to the input / output controller 2220 via an input / output chip 2240.

[0140] The CPU 2212 operates according to programs stored in the ROM 2230 and RAM 2214, thereby controlling each unit. The graphics controller 2216 acquires image data generated by the CPU 2212 into a frame buffer or the like provided in the RAM 2214 or into the graphics controller 2216 itself, and causes the image data to be displayed on the display device 2218.

[0141] The communications interface 2222 communicates with other electronic devices via a network. The hard disk drive 2224 stores programs and data used by the CPU 2212 in the computer 2200. The DVD-ROM drive 2226 reads programs or data from the DVD-ROM 2201 and provides the programs or data to the hard disk drive 2224 via the RAM 2214. The IC card drive reads programs and data from an IC card and / or writes programs and data to an IC card.

[0142] The ROM 2230 stores therein a boot program or the like that is executed by the computer 2200 upon activation, and / or programs that depend on the hardware of the computer 2200. The input / output chip 2240 may also connect various input / output units to the input / output controller 2220 via a parallel port, a serial port, a keyboard port, a mouse port, etc.

[0143] The programs are provided by a computer-readable medium such as a DVD-ROM 2201 or an IC card. The programs are read from the computer-readable medium, installed in the hard disk drive 2224, RAM 2214, or ROM 2230, which are also examples of computer-readable media, and executed by the CPU 2212. Information processing described in these programs is read by the computer 2200, and brings about cooperation between the programs and the various types of hardware resources described above. An apparatus or method may be configured by realizing information manipulation or processing in accordance with the use of the computer 2200.

[0144] For example, when communication is performed between the computer 2200 and an external device, the CPU 2212 may execute a communication program loaded into the RAM 2214 and instruct the communication interface 2222 to perform communication processing based on the processing described in the communication program. Under the control of the CPU 2212, the communication interface 2222 reads transmission data stored in a transmission buffer processing area provided in the RAM 2214, the hard disk drive 2224, the DVD-ROM 2201, or a recording medium such as an IC card, and transmits the read transmission data to the network, or writes reception data received from the network to a reception buffer processing area or the like provided on the recording medium.

[0145] The CPU 2212 may cause all or a necessary portion of a file or database stored on an external recording medium such as the hard disk drive 2224, the DVD-ROM drive 2226 (DVD-ROM 2201), an IC card, etc. to be read into the RAM 2214, and may perform various types of processing on the data on the RAM 2214. The CPU 2212 then writes back the processed data to the external recording medium.

[0146] Various types of information, such as various types of programs, data, tables, and databases, may be stored on the recording medium and may undergo information processing. The CPU 2212 may perform various types of processing on data read from the RAM 2214, including various types of operations, information processing, conditional judgment, conditional branching, unconditional branching, information search / replacement, etc., as described throughout this disclosure and specified by the instruction sequences of the programs, and write the results back to the RAM 2214. The CPU 2212 may also search for information in a file, database, etc. on the recording medium. For example, if multiple entries each having an attribute value of a first attribute associated with an attribute value of a second attribute are stored on the recording medium, the CPU 2212 may search for an entry that matches a condition specified by the attribute value of the first attribute from among the multiple entries, read the attribute value of the second attribute stored in the entry, and thereby obtain the attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

[0147] The above-described programs or software modules may be stored in a computer-readable medium on or near the computer 2200. A recording medium such as a hard disk or RAM provided in a server system connected to a dedicated communication network or the Internet can also be used as a computer-readable medium, thereby providing the programs to the computer 2200 via the network.

[0148] Although the present invention has been described above using embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments. It will be apparent to those skilled in the art that various modifications and improvements can be made to the above embodiments. It is clear from the claims that such modifications and improvements can also be included within the technical scope of the present invention.

[0149] It should be noted that the execution order of each process, such as operations, procedures, steps, and stages, in the devices, systems, programs, and methods shown in the claims, specifications, and drawings is not specifically stated as "before," "prior to," etc., and that the processes can be performed in any order unless the output of a previous process is used in a subsequent process. Even if the operational flow in the claims, specifications, and drawings is described using "first," "next," etc. for convenience, this does not mean that the processes must be performed in this order. [Explanation of symbols]

[0150] 100 Prediction System 110 databases 120 Data Generation Unit 130 Model Generation Unit 132 Classification Department 134 Classification Learning Unit 136 Regression Learning Unit 138 Regression coefficient acquisition unit 140 Prediction Department 150 Performance output section 200 Manufacturing equipment 210 Mixing section 220 Generation part 300 Composition Group Data 310 Composition information 320 performance value 330 Performance classification 400 Composition group data after classification 410 Composition information 420 performance value 430 Performance classification 500 graphs 510 Composition Group Data 520 Composition Group Data 530 Regression Models 540 Regression Models 550 typical value 560 typical value 610 Composition information 620 Predicted performance values ​​for each performance category 630 Overall predicted performance value 2200 Computer 2201 DVD-ROM 2210 host controller 2212 CPU 2214 RAM 2216 Graphics Controller 2218 Display Device 2220 Input / Output Controller 2222 communication interface 2224 hard disk drive 2226 DVD-ROM drive 2230 ROM 2240 I / O chip 2242 keyboard

Claims

1. 1. A prediction system for predicting the performance of a composition, comprising: a model generation unit that receives composition group data including a plurality of pairs of composition information, a performance value of the composition, and a performance classification based on the performance of the composition, and generates a plurality of prediction models that predict the performance value from the composition information based on the performance classification; Prediction system.

2. The model generation unit a classification unit that classifies composition group data based on the performance classification; a classification learning unit that generates a prediction model that predicts the performance value from the composition information for each of the composition group data classified based on the performance classification; The prediction system of claim 1 , comprising:

3. The model generation unit a regression learning unit that generates a regression model that predicts the performance value from the composition information based on the performance classification, The prediction system of claim 1 .

4. the model generation unit includes a regression coefficient acquisition unit that acquires regression coefficients in the regression model based on the performance classification; The prediction system of claim 3 .

5. the regression learning unit generates a regression model including a probability distribution of regression coefficients based on the performance classification; the regression coefficient acquisition unit acquires a representative value of the regression coefficient by sampling a regression model including a probability distribution of the regression coefficient. The prediction system of claim 4 .

6. the regression coefficient acquisition unit performs the sampling using a Markov Chain Monte Carlo method (MCMC). The prediction system of claim 5 .

7. The composition group data is Further includes unclassified data including composition information of the composition and performance values ​​of the composition, and not including performance classifications; The prediction system of claim 3 .

8. The regression learning unit applying Gaussian mixture regression to the composition group data to generate a regression model that predicts the performance value from the composition information based on the unclassified data; The prediction system of claim 7 .

9. The performance values ​​are within the range of products that can be produced with the composition. The prediction system of claim 1 .

10. The performance value is the minimum size of a product that can be produced using the composition. The prediction system of claim 9 .

11. The performance classification is a classification based on the appearance of a product produced using the composition. The prediction system of claim 1 .

12. The system further includes a data generating unit that generates the composition group data by assigning a performance classification to the composition based on an appearance image of a product manufactured using the composition. The prediction system of claim 11.

13. the composition is a resin composition, The composition information includes information on the polymer resin, monomers, and other components in the resin composition. The prediction system of claim 1 .

14. a prediction unit that inputs composition information of a composition to be predicted into the prediction model and outputs the performance value based on the performance classification of the composition; The prediction system of claim 1 .

15. a performance output unit that outputs the worst or best performance value among the performance values ​​based on the performance classification as an overall performance value of the composition; The prediction system of claim 14.

16. 1. A prediction system for predicting the performance of a composition, comprising: a model generation unit that receives composition group data including a plurality of pairs of composition information and performance values ​​of the compositions and generates a plurality of prediction models that predict the performance values ​​from the composition information; Prediction system.

17. a prediction unit that inputs composition information of a composition to be predicted into a prediction model and outputs a performance value based on a performance classification that is a classification based on the performance of the composition; a performance output unit that outputs the worst or best performance value among the performance values ​​based on the performance classification as an overall performance value of the composition; A prediction system comprising:

18. The prediction system according to any one of claims 1 to 15, a model generation step of inputting composition group data including a plurality of pairs of composition information, a performance value of the composition, and a performance classification based on the performance of the composition, and generating a plurality of prediction models that predict the performance value from the composition information based on the performance classification; A forecasting method that includes:

19. The prediction system of claim 16 comprises: a model generation step of inputting composition group data including a plurality of pairs of composition information and performance values ​​of the compositions, and generating a plurality of prediction models that predict the performance values ​​from the composition information; A prediction method comprising:

20. The prediction system of claim 17, a prediction step of inputting composition information of a composition to be predicted into a prediction model and outputting a performance value based on a performance classification that is a classification based on the performance of the composition; a performance output step of outputting the worst or best performance value among the performance values ​​based on the performance classification as an overall performance value of the composition; A prediction method comprising:

21. The method is executed by a computer, causing the computer to: The prediction system according to any one of claims 1 to 17 is operated as follows: program.

Images