Information processing device, information processing system, information processing method, and information processing program

Abstract

This provides a technique for setting appropriate upper and lower limits for data containing multiple features. [Solution] The information processing device comprises an acquisition unit that acquires target data, a designation unit that designates one or more explanatory variables and one or more target variables from a plurality of feature quantities included in the target data, and a derivation unit that derives a first linear combination defining the upper limit of the target variable, which is a first linear combination of one or more explanatory variables, and a second linear combination defining the lower limit of the target variable, which is a second linear combination of one or more explanatory variables.

Description

[Technical Field]

[0001] This disclosure relates to an information processing device, an information processing system, an information processing method, and an information processing program. [Background technology]

[0002] Techniques related to upper and lower limits on data are known. For example, Patent Document 1 discloses a production support system that, when an element value related to the production elements of a product exceeds an acceptable range, resets the product quality to an acceptable range that satisfies a predetermined quality. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2021-56671 [Overview of the Initiative] [Problems that the invention aims to solve]

[0004] The production support system described in Patent Document 1 does not consider the tolerance range when there are multiple production elements. In other words, the production support system described in Patent Document 1 does not consider setting an tolerance range for element values ​​from production elements that include multiple features. On the other hand, in fields such as business optimization, there is a practical need to appropriately set or detect upper and lower limits from data that includes multiple features.

[0005] This disclosure has been made in view of the above-mentioned issues, and one exemplary purpose is to provide a technique for setting appropriate upper and lower limits for data containing multiple features. [Means for solving the problem]

[0006] An information processing device relating to an illustrative aspect of this disclosure includes: acquisition means for acquiring target data including a plurality of features; designation means for specifying one or more explanatory variables and one or more target variables from the plurality of features included in the target data; and derivation means for deriving a first linear combination defining the upper limit of the target variable, which is a first linear combination of the one or more explanatory variables, and a second linear combination defining the lower limit of the target variable, which is a second linear combination of the one or more explanatory variables.

[0007] An illustrative aspect of the present disclosure relates to an information processing system comprising a first information processing device and a second information processing device, wherein the first information processing device includes acquisition means for acquiring target data including a plurality of features, selection means for selecting one or more explanatory variables and one or more target variables from the plurality of features included in the target data, and derivation means for deriving a first linear combination defining an upper limit of the target variable, which is a first linear combination of the one or more explanatory variables, and a second linear combination defining a lower limit of the target variable, which is a second linear combination of the one or more explanatory variables, and the second information processing device includes optimization means for performing an optimization process that refers to at least a portion of the target data under constraints defined using at least one of the first linear combination and the second linear combination.

[0008] An information processing method relating to an illustrative aspect of this disclosure includes: an acquisition process in which at least one processor acquires target data including a plurality of features; a designation process in which the at least one processor designates one or more explanatory variables and one or more target variables from the plurality of features included in the target data; and a derivation process in which the at least one processor derives a first linear combination defining an upper limit of the target variable, which is a first linear combination of the one or more explanatory variables, and a second linear combination defining a lower limit of the target variable, which is a second linear combination of the one or more explanatory variables.

[0009] An illustrative aspect of the present disclosure relates to an information processing program, which is a program that causes a computer to function as an information processing device, and the computer functions as an acquisition means for acquiring target data including a plurality of features, a designation means for specifying one or more explanatory variables and one or more target variables from the plurality of features included in the target data, and a derivation means for deriving a first linear combination that defines the upper limit of the target variable, which is a first linear combination of the one or more explanatory variables, and a second linear combination that defines the lower limit of the target variable, which is a second linear combination of the one or more explanatory variables. [Effects of the Invention]

[0010] One illustrative aspect of this disclosure provides a technique for setting appropriate upper and lower limits for data containing multiple features, which is one exemplary effect. [Brief explanation of the drawing]

[0011] [Figure 1] This is a block diagram showing the configuration of the information processing device related to this disclosure. [Figure 2] This is a flowchart showing the flow of the information processing method related to this disclosure. [Figure 3] This is a block diagram showing the configuration of the information processing system related to this disclosure. [Figure 4] This is a block diagram showing the configuration of the information processing system related to this disclosure. [Figure 5] This figure schematically shows the output of each endpoint model in the LPV model relating to this disclosure, and the internal division ratio parameter multiplied by each output. [Figure 6] This figure shows an example of the processing flow in the information processing device related to this disclosure. [Figure 7] This document shows an example of a graph displayed by the output unit related to this disclosure via the input / output unit. [Figure 8] This figure shows another example of the processing flow in the information processing device related to this disclosure. [Figure 9]This figure shows an example of a graph displayed by the output unit via the input / output unit and explanatory information displayed by the generation unit via the input / output unit, according to this disclosure. [Figure 10] This is a block diagram showing the configuration of the information processing system related to this disclosure. [Figure 11] This is a block diagram showing the configuration of the information processing device, the first information processing device, the second information processing device, and the computer that functions as an optimization device, according to this disclosure. [Modes for carrying out the invention]

[0012] The following are examples of embodiments of the present invention. However, the present invention is not limited to the exemplary embodiments shown below, and various modifications are possible within the scope of the claims. For example, embodiments obtained by appropriately combining some or all of the technologies (things or methods) employed in each of the exemplary embodiments shown below may also be included in the scope of the present invention. Furthermore, embodiments obtained by appropriately omitting some of the technologies employed in each of the exemplary embodiments shown below may also be included in the scope of the present invention. In addition, the effects mentioned in each of the exemplary embodiments shown below are examples of effects that can be expected in that exemplary embodiment and do not define the scope of the present invention. That is, embodiments that do not produce the effects mentioned in each of the exemplary embodiments shown below may also be included in the scope of the present invention.

[0013] [First Exemplary Embodiment] A first exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. This exemplary embodiment is the basic form for each of the exemplary embodiments described later. The scope of application of each technology adopted in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology adopted in this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems occur. Furthermore, each technology shown in the drawings referenced to explain this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems occur.

[0014] (Configuration of Information Processing Device 1) The configuration of the information processing device 1 will be described with reference to Figure 1. Figure 1 is a block diagram showing the configuration of the information processing device 1. As shown in Figure 1, the information processing device 1 includes an acquisition unit 11, a designation unit 12, and a derivation unit 13. In this exemplary embodiment, the acquisition unit 11, the designation unit 12, and the derivation unit 13 realize the acquisition means, the designation means, and the derivation means, respectively.

[0015] (Acquisition part 11) The acquisition unit 11 acquires target data that includes multiple feature quantities. The acquisition unit 11 supplies the acquired target data to the designation unit 12 and the derivation unit 13.

[0016] (Specification part 12) The designation unit 12 selects one or more explanatory variables and one or more target variables from multiple features included in the target data. The designation unit 12 supplies the derivation unit 13 with information indicating the selected one or more explanatory variables and one or more target variables.

[0017] (Derivation part 13) The derivation unit 13 derives a first linear combination that defines the upper limit of the dependent variable, which is a first linear combination of one or more explanatory variables, and a second linear combination that defines the lower limit of the dependent variable, which is a second linear combination of one or more explanatory variables.

[0018] The derivation unit 13 trains multiple regression models, each associated with one of the multiple ratio parameters defined by the hidden variables, by referring to at least a portion of the target data, and derives the first linear sum and the second linear sum using the multiple regression models.

[0019] (Effects of Information Processing Device 1) As described above, the information processing device 1 employs a configuration comprising: an acquisition unit 11 that acquires target data containing multiple feature quantities; a designation unit 12 that designates one or more explanatory variables and one or more target variables from the multiple feature quantities contained in the target data; and a derivation unit 13 that derives a first linear combination that defines the upper limit of the target variable, which is a first linear combination of one or more explanatory variables, and a second linear combination that defines the lower limit of the target variable, which is a second linear combination of one or more explanatory variables.

[0020] Therefore, the information processing device 1 has the effect of being able to set appropriate upper and lower limits for data containing multiple features.

[0021] (Information processing method S1 flow) The flow of the information processing method S1 will be explained with reference to Figure 2. Figure 2 is a flowchart showing the flow of the information processing method S1. As shown in Figure 2, the information processing method S1 includes an acquisition process S11, a specification process S12, and a derivation process S13.

[0022] (Acquisition process S11) In the acquisition process S11, the acquisition unit 11 acquires target data containing multiple feature quantities. The acquisition unit 11 supplies the acquired target data to the designation unit 12 and the derivation unit 13.

[0023] (Specified processing S12) In the specification process S12, the specification unit 12 selects one or more explanatory variables and one or more target variables from a plurality of features included in the target data. The specification unit 12 supplies the derivation unit 13 with information indicating the selected one or more explanatory variables and one or more target variables.

[0024] (Derivation process S13) In the derivation process S13, the derivation unit 13 derives a first linear combination that defines the upper limit of the dependent variable, which is a first linear combination of one or more explanatory variables, and a second linear combination that defines the lower limit of the dependent variable, which is a second linear combination of one or more explanatory variables.

[0025] (Effects of information processing method S1) As described above, the information processing method S1 employs a configuration in which the acquisition unit 11 performs an acquisition process S11 to acquire target data containing multiple feature quantities, the designation unit 12 performs a designation process S12 to designate one or more explanatory variables and one or more target variables from the multiple feature quantities contained in the target data, and the derivation unit 13 performs a derivation process S13 to derive a first linear combination that defines the upper limit of the target variable, which is a first linear combination of one or more explanatory variables, and a second linear combination that defines the lower limit of the target variable, which is a second linear combination of one or more explanatory variables. Therefore, the same effects as those of the information processing device 1 described above can be obtained with the information processing method S1.

[0026] (Configuration of Information Processing System 100) The configuration of the information processing system 100 will be explained with reference to Figure 3. Figure 3 is a block diagram showing the configuration of the information processing system 100. As shown in Figure 3, the information processing system 100 includes a first information processing device 1 and a second information processing device 2.

[0027] (First information processing device 1) As shown in Figure 3, the first information processing device 1 includes an acquisition unit 11, a designation unit 12, and a derivation unit 13. In this exemplary embodiment, the acquisition unit 11, the designation unit 12, and the derivation unit 13 implement the acquisition means, the designation means, and the derivation means, respectively.

[0028] (Acquisition part 11) The acquisition unit 11 acquires target data that includes multiple feature quantities. The acquisition unit 11 supplies the acquired target data to the designation unit 12 and the derivation unit 13.

[0029] (Specification part 12) The designation unit 12 selects one or more explanatory variables and one or more target variables from multiple features included in the target data. The designation unit 12 supplies the derivation unit 13 with information indicating the selected one or more explanatory variables and one or more target variables.

[0030] (Derivation part 13) The derivation unit 13 derives a first linear combination that defines the upper limit of the dependent variable, which is a first linear combination of one or more explanatory variables, and a second linear combination that defines the lower limit of the dependent variable, which is a second linear combination of one or more explanatory variables.

[0031] (Second information processing device 2) As shown in Figure 3, the second information processing device 2 includes an optimization unit 21. In this exemplary embodiment, the optimization unit 21 implements the optimal means.

[0032] (Optimization unit 21) The optimization unit 21 performs an optimization process that references at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination.

[0033] (Effects of Information Processing System 100) As described above, the information processing system 100 employs a configuration that includes a first information processing device 1 and a second information processing device 2.

[0034] The first information processing device 1 employs a configuration comprising: an acquisition unit 11 that acquires target data containing multiple features; a designation unit 12 that designates one or more explanatory variables and one or more target variables from the multiple features contained in the target data; and a derivation unit 13 that derives a first linear combination defining the upper limit of the target variable, which is a first linear combination of one or more explanatory variables, and a second linear combination defining the lower limit of the target variable, which is a second linear combination of one or more explanatory variables.

[0035] The second information processing device 2 employs a configuration that includes an optimization unit 21 that performs optimization processing by referencing at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination.

[0036] Therefore, the information processing system 100 can achieve the same effects as the information processing device 1 described above.

[0037] [Second exemplary embodiment] A second exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same function as those described in the above-described exemplary embodiment are denoted by the same reference numerals, and their descriptions are omitted as appropriate. The scope of application of each technology adopted in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology adopted in this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems arise. Furthermore, each technology shown in the drawings referenced to describe this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems arise.

[0038] (Configuration and overview of information processing system 100A) The overview of the information processing system 100A will be explained with reference to Figure 4. Figure 4 is a block diagram showing the configuration of the information processing system 100A. As shown in Figure 4, the information processing system 100A includes an information processing device 1A and an optimization device 60.

[0039] In the information processing system 100A, the information processing device 1A and the optimization device 60 are connected in a communicative manner. For example, as shown in Figure 4, the information processing device 1A and the optimization device 60 are connected in a communicative manner via a network N. The specific configuration of network N is not particularly limited, but as an example, a wireless LAN (Local Area Network), wired LAN, WAN (Wide Area Network), public telephone network, mobile data communication network, or a combination of these networks can be used.

[0040] In the information processing system 100A, the information processing device 1A selects one or more explanatory variables and one or more target variables from a plurality of features contained in the target data TD. The information processing device 1A then derives a first linear combination LC1 of one or more explanatory variables, which defines the upper limit of the target variable. The information processing device 1A also derives a second linear combination LC2 of one or more explanatory variables, which defines the lower limit of the target variable. The information processing device 1A then outputs the derived first linear combination LC1 and second linear combination LC2 to the optimization device 60.

[0041] Furthermore, in the information processing system 100A, the optimization device 60 performs optimization processing by referring to at least a portion of the target data TD, under constraints defined using at least one of the first linear sum LC1 and the second linear sum LC2 output from the information processing device 1A.

[0042] As an example, the information processing device 1A acquires log data for each of the multiple tasks performed by each of the multiple workers (for example, the time worker A spent performing tasks X and Y, the time worker B spent performing tasks X and Z, etc.) as target data TD. The information processing device 1A then outputs a first linear sum LC1 indicating the upper limit of the task time and a second linear sum LC2 indicating the lower limit of the task time to the optimization device 60 for each combination of worker and task.

[0043] The optimization device 60 obtains a first linear sum LC1 indicating the upper limit of the work time and a second linear sum LC2 indicating the lower limit of the work time for each combination of worker and work from the information processing device 1A, and then performs optimization processing of the combination of worker and work time. For example, the optimization device 60 determines a combination of worker and work time to perform a predetermined work so that the predetermined work is completed within a predetermined time.

[0044] Furthermore, the information processing system 100A may be configured in which optimization processing is performed under constraints.

[0045] As an example, the optimization device 60 may be configured to perform an optimization process that references at least a portion of the target data TD under constraints defined using at least one of the first linear sum LC1 and the second linear sum LC2.

[0046] As another example, the optimization device 60 may perform an optimization process that references at least a portion of the target data TD without any constraints. In this case, if the optimization process by the optimization device 60 results in a combinatorial explosion (i.e., no solution is found), the optimization device 60 may be configured to derive constraints and perform the optimization process under those constraints.

[0047] (An example of a processing algorithm used in information processing system 100A) An example of a processing algorithm used in the information processing system 100A according to this exemplary embodiment will be described. The inventors are investigating the Linear Parameter-Varying (LPV) Model as a modeling method for systems with variation. In this LPV model, as an example, the internal state variables x k , and output state variables (output state variables) y k This is updated and calculated using the following formulas (1A) and (1B).

[0048]

number

[0049] Here, A (i) B (i) μ is a matrix that represents each state-space model (also called each endpoint model) that is identified by index i, and μ (i) k μ is a parameter that defines the internal division ratio (weight) of each model. (i) k This is sometimes called the internal division ratio parameter, the weight parameter, or the scheduling parameter. Also, in the above LPV model, uk is the input amount (input variable) as an example, and C and D are x k and u k respectively, which are output matrices calculated for x and u. k is the index assigned to each state variable and is, for example, time.

[0050] FIG. 5 schematically shows the outputs of each endpoint model in the above LPV model (1-st SS model to 5-th SS model in FIG. 5) and the interpolation ratio parameter μ multiplied by each output. (i) k As shown in FIG. 5, for each of the outputs of a plurality of endpoint models at the k-th step (A (i) x k +B (i) u k ) (i = 1 to 5) , each interpolation ratio parameter μ (i) k (i = 1 to 5) is multiplied, and x at the (k + 1)-th step k+1 is calculated.

[0051] Such an LPV model has an aspect that it is suitable for modeling a system with variations. On the other hand, there is a problem that it is difficult to apply to a system in which the value of the interpolation ratio parameter μ (i) k is not clear.

[0052] The inventor · treats the above interpolation ratio parameter μ (i) k as a hidden variable (posterior probability) z k and · applies the learning method of the hidden variable model in machine learning · calculates the above interpolation ratio parameter μ (i) k as the expected value of the hidden variable z k and has obtained the knowledge that learning of the LPV model can be realized even when the interpolation ratio parameter is unknown. More specifically, the inventor has found that the interpolation ratio parameter μ​(i) k The L2PV model (Latent Linear Parameter-Varying model) is defined by the following equations (2A) to (2C), which introduce as a hidden variable.

[0053]

number

[0054] The following equations (3A) to (3E)

[0055]

number

[0056] By rewriting the regression model into the form defined by (L2PV regression model), the internal ratio parameter μ (i) k I came up with the idea of ​​making that the subject of study.

[0057] The processing in the information processing system 100A described below is based on the formulation described above and is a process based on the inventor's unique perspective.

[0058] (Configuration of Information Processing Device 1A) The configuration of the information processing device 1A will be explained again with reference to Figure 4. As shown in Figure 4, the information processing device 1A comprises a control unit 10A, a storage unit 15A, a communication unit 16A, and an input / output unit 17A.

[0059] (Storage section 15A) First, let's describe the various types of data (information) stored in the storage unit 15A. The storage unit 15A stores data that the control unit 10A references. Examples of the storage unit 15A include, but are not limited to, flash memory, HDD (Hard Disk Drive), SSD (Solid State Drive), or a combination thereof.

[0060] Examples of data stored in the memory unit 15A include, but are not limited to, the target data TD, internal ratio parameter RP, regression coefficient RC, distribution information DI, learning result LR, first linear sum LC1, second linear sum LC2, and explanatory information EI, as shown in Figure 5.

[0061] The target data TD contains multiple features and is used for learning processing in the information processing device 1A. The target data TD is a state variable (~x k ) and state variables (~y k The set of ) is expressed by the following equation (4). In this specification, the state variable x k ,~x k ,y k ,~y k These are sometimes called features. Also, the state variable x k ,~x k These are sometimes called explanatory variables, or state variables y k ,~y k This is sometimes referred to as the dependent variable. Furthermore, if the dependent variable is the object of derivation, it may also be called the predicted value. These specific terms are not limited to those described herein.

[0062]

number

[0063] The internal ratio parameter RP is a parameter that defines the relative weights of multiple state-space models in an LPV model, and is also called the scheduling parameter. Furthermore, the internal ratio parameter RP is sometimes referred to as the weight parameter RP or the scheduling parameter RP. As an example, the internal ratio parameter RP is given by the following equation (5) for each of the m models (model 1 to model m).

[0064]

number

[0065] Here, k is an index similar to the index assigned to each state variable described above, and N represents the dimension of each state variable (the number of samples for each state variable). Also, the index (i) related to the model is not explicitly shown in the above expression. This is because the internal division ratio parameter RP is an internal division ratio parameter vector composed of components corresponding to models 1 to m for each k. μ k =(μ k (1) , μ k (2) , , , μ k (m) ) It may also be interpreted as being expressed in this way. Thus, the internal division ratio parameter RP can also be expressed as an internal division ratio parameter vector or an internal division ratio parameter matrix. In this disclosure, the case where m=2 is described as an example.

[0066] Furthermore, the internal ratio parameter μ for a certain model j k (j) This is the N-dimensional target data x k It can also be expressed as the components of an N-dimensional vector having components corresponding to each of (k=1 to N). More specifically, it is the j-th internal division ratio parameter μ. k (j) This is the N-dimensional target data x k (μ1) corresponds to each component (k=1~N) (j) , μ2 (j) , , , μ N (j) These are the components of an N-dimensional vector having ).

[0067] This disclosure is not limited to internal ratio parameters RP, but may also refer to external ratio parameters. Hereinafter, internal ratio parameters RP may be referred to as ratio parameters RP.

[0068] The regression coefficient RC is a coefficient in the L2PV regression model. The regression coefficient RC is expressed by the following equation (6).

[0069]

number

[0070] The distribution information DI includes the covariance matrix Φ of the prior distribution of the hidden variable, the covariance parameter η of the prior distribution of the hidden variable, and the covariance parameter Ψ of the posterior distribution of the hidden variable.

[0071] Hidden variable z k Prior distribution p(z k ) is expressed as equation (7) below.

[0072]

number

[0073] Hidden variable z k The covariance parameter η of the prior distribution is, in other words, the model likelihood p(~y) expressed by the following equation (8). k |z k ,~x k This is the covariance parameter η of ( ,W,η).

[0074]

number

[0075] Hidden variable z k The posterior distribution p(z) k |~y k ,~x k ,W,η) is expressed as equation (9) below.

[0076]

number

[0077] In the above formula, the N of the calligraphy typeface on the right-hand side represents a normal distribution. However, this does not mean that the distribution examples in this exemplary embodiment are limited to the normal distribution. For example, the hidden variable z kThe Dirichlet distribution may be used as the posterior distribution.

[0078] Furthermore, as will be described later, in the processing by the information processing device 1A, the hidden variable z k The posterior distribution p(z) k |~y k ,~x k The expression ,W,η) is given by the following constraint (condition) in equation (10).

[0079]

number

[0080] Therefore, the hidden variable z k Even when a normal distribution is used as the posterior distribution, suitable calculations can be performed.

[0081] The learning result LR includes the calculated internal ratio parameter RP and the calculated regression coefficient RC.

[0082] The first linear sum LC1 is equal to the internal ratio parameter μ with respect to Model 1. k (1) It is a linear combination specified by the corresponding regression coefficient RC. In other words, the first linear combination LC1 is a linear combination that defines the upper limit of the dependent variable and is a linear combination of one or more explanatory variables. The first linear combination LC1 is expressed as equation (11) below, where y is the dependent variable and f={f1,f2,...} is the explanatory variable.

[0083]

number

[0084] The second linear sum LC2 is equal to the internal ratio parameter μ for Model 2. k (2)It is a linear combination specified by the corresponding regression coefficient RC. In other words, the second linear combination LC2 is a linear combination that defines the lower bound of the dependent variable and is a linear combination of one or more explanatory variables. The second linear combination LC2 is expressed as equation (12) below, where y is the dependent variable and f={f1,f2,...} is the explanatory variable.

[0085]

number

[0086] Explanatory information EI is information about explanatory variables corresponding to one or more coefficients included in at least one of the first linear combination LC1 and the second linear combination LC2. For example, if the first linear combination LC1 and the second linear combination LC2 are expressed as equations (11) and (12) above, respectively, then explanatory information EI is information about explanatory variables f corresponding to at least one of the coefficients a={a1, a2, ...} and coefficients c={c1, c2, ...}. In other words, explanatory information EI can also be expressed as information that quantifies the correspondence between the dependent variable y and the explanatory variable f corresponding to that dependent variable y.

[0087] (Communications Section 16A) The communication unit 16A is an interface for sending and receiving data over a network. Examples of the communication unit 16A include, but are not limited to, communication chips in various communication standards such as Ethernet (registered trademark), Wi-Fi (Wireless Fidelity, registered trademark), and wireless communication standards for mobile data communication networks, as well as USB-compliant connectors.

[0088] (Input / output section 17A) The input / output unit 17A is an interface that accepts data input and outputs data. Examples of input / output units 17A include, but are not limited to, microphones, cameras, eye-tracking devices, keyboards, touchpads, speakers, and liquid crystal displays.

[0089] (Control Unit 10A) The control unit 10A controls each component of the information processing device 1A. Furthermore, as shown in Figure 4, the control unit 10A includes an acquisition unit 11, a designation unit 12, a derivation unit 13, and a generation unit 14. In this exemplary embodiment, the acquisition unit 11, designation unit 12, derivation unit 13, and generation unit 14 implement the acquisition means, designation means, derivation means, and generation means, respectively.

[0090] The acquisition unit 11 acquires data via the communication unit 16A or the input / output unit 17A. An example of the data acquired by the acquisition unit 11 is the target data TD and the hidden variable z. k Information regarding the prior distribution is provided. The acquisition unit 11 stores the acquired data in the storage unit 15A.

[0091] The designation unit 12 specifies one or more explanatory variables and one or more target variables from a set of features included in the target data TD. For example, the designation unit 12 specifies features x1 and x2 included in the set of features (~xk) as explanatory variables, and features y1 and y2 included in the set of features (~yk) as target variables. As another example, the designation unit 12 specifies the sum of features x1 and x2 included in the set of features (~xk) as an explanatory variable. As yet another example, it specifies the difference between features y1 and y2 included in the set of features (~yk) as the target variable. The designation unit 12 may also be described as a selection means that selects one or more explanatory variables and one or more target variables from a set of features included in the target data TD.

[0092] The derivation unit 13 derives a first linear sum LC1 and a second linear sum LC2. As an example, the derivation unit 13 trains multiple regression models, each associated with a plurality of ratio parameters RP defined by a hidden variable, by referring to at least a portion of the target data TD, and derives the first linear sum LC1 and the second linear sum LC2 using the plurality of regression models. In this disclosure, as described above, the derivation unit 13 trains two regression models by referring to at least a portion of the target data TD.

[0093] Furthermore, as shown in Figure 4, the derivation unit 13 includes a regression coefficient calculation unit 132, a covariance calculation unit 133, an internal ratio parameter calculation unit 134, an output unit 135, an initial value determination unit 136, and a convergence determination unit 137. In this exemplary embodiment, the regression coefficient calculation unit 132, the covariance calculation unit 133, and the internal ratio parameter calculation unit 134 implement the regression coefficient calculation means, the covariance calculation means, and the ratio parameter calculation means, respectively.

[0094] The regression coefficient calculation unit 132 calculates the regression coefficient RC for each of the multiple target models by referring to the internal ratio parameter RP, which defines the internal ratio of multiple target models, and the target data TD. For example, the internal ratio parameter RP referred to by the regression coefficient calculation unit 132 is the initial value of the internal ratio parameter RP determined by the initial value determination unit 136, which will be described later. As another example, the internal ratio parameter RP referred to by the regression coefficient calculation unit 132 is the internal ratio parameter RP calculated by the internal ratio parameter calculation unit 134, which will be described later. The regression coefficient calculation unit 132 stores the calculated regression coefficient RC in the storage unit 15A. As described above, the internal ratio parameter RP may also be an external ratio parameter. Furthermore, the internal ratio parameter RP defines the internal ratio of two target models.

[0095] The covariance calculation unit 133 calculates the target data TD, the internal ratio parameter RP, the regression coefficient RC, and the hidden variable z. k Referencing the covariance matrix Φ of the prior distribution, the hidden variable z k The covariance parameter η of the prior distribution and the hidden variable z k The covariance matrix Ψ of the posterior distribution is calculated. The covariance calculation unit 133 calculates the hidden variable z k The covariance parameter η of the prior distribution and the hidden variable z k The covariance matrix Ψ of the posterior distribution is stored in the memory unit 15A as distribution information DI.

[0096] The internal ratio parameter calculation unit 134 calculates the target data TD, the regression coefficient RC, and the hidden variable z kThe internal ratio parameter RP is calculated by referring to the covariance matrix Ψ of the posterior distribution. The internal ratio parameter calculation unit 134 stores the calculated internal ratio parameter RP in the storage unit 15A. The internal ratio parameter calculation unit 134 may also calculate the external ratio parameter.

[0097] With the above configuration, the derivation unit 13 can suitably derive the first linear sum LC1 and the second linear sum LC2 using the L2PV regression model defined by the L2PV model described above.

[0098] The output unit 135 outputs data via the communication unit 16A and the input / output unit 17A. As an example, the output unit 135 outputs a first linear sum LC1 and a second linear sum LC2. As another example, the output unit 135 outputs explanatory information EI.

[0099] The initial value determination unit 136 determines the initial value of the internal ratio parameter RP, which is referenced by the regression coefficient calculation unit 132. The initial value determination unit 136 stores the determined initial value of the internal ratio parameter RP in the storage unit 15A.

[0100] The convergence determination unit 137 determines whether the calculations relating to the internal division ratio parameter RP have converged. The convergence determination unit 137 supplies the determination result to the output unit 135.

[0101] The generation unit 14 generates explanatory information EI. The generation unit 14 also outputs the generated explanatory information EI. For example, if one or more target variables include one or more indicators related to the work, and one or more explanatory variables include features related to the workers performing the work, the generation unit 14 generates information about the workers' skill level as explanatory information EI.

[0102] Furthermore, the generation unit 14 may be described as generating explanatory information EI relating to the explanatory variables corresponding to the one or more coefficients by referring to one or more coefficients included in at least one of the following: a first linear combination LC1 that defines the upper limit of the dependent variable, which is a first linear combination LC1 of one or more explanatory variables, and a second linear combination LC2 that defines the lower limit of the dependent variable, which is a second linear combination LC2 of one or more explanatory variables.

[0103] (Configuration of the optimization device 60) The configuration of the optimization device 60 will be explained again with reference to Figure 4. As shown in Figure 4, the optimization device 60 includes a control unit 61 and a communication unit 62. The communication unit 62 has the same functions as the communication unit 16A described above, so its explanation will be omitted.

[0104] The control unit 61 controls each component of the optimization device 60. The control unit 61 also includes an optimization unit 63, as shown in Figure 4. In this exemplary embodiment, the optimization unit 63 implements the optimization means.

[0105] The optimization unit 63 performs an optimization process that references at least a portion of the target data TD, under constraints defined using at least one of the first linear sum LC1 and the second linear sum LC2.

[0106] Furthermore, the optimization unit 63 may perform an optimization process that refers to at least a portion of the target data TD without any constraints. In this case, if a combinatorial explosion occurs during the optimization process (i.e., no solution is found), the optimization unit 63 will derive constraints by referring to at least one of the first linear combination LC1 and the second linear combination LC2, and will perform the optimization process under those constraints.

[0107] (Example 1 of the processing flow in information processing device 1A) FIG. 6 is a diagram showing an example of the processing flow in the information processing apparatus 1A according to this exemplary embodiment. Note that the processing example described below can also be regarded as a variational Bayes EM algorithm, but this does not limit this exemplary embodiment. Further, the processing example described below can be regarded as a process of updating each parameter so as to maximize the variational lower bound (VLB) J obtained by the following formula (13).

[0108]

Number

[0109] Also, the processing example described below can also be expressed as an algorithm for solving the maximum likelihood problem defined by the model likelihood p in the following formula (14).

[0110]

Number

[0111] (Step S11: Acquisition process) In step S11, the acquisition unit 11 acquires the target data TD. Here, as described above, the target data TD is data used for the learning process in the information processing apparatus 1A. Since the details of the target data TD have been described, the description is omitted here.

[0112] Also, in step S11, the acquisition unit 11 further acquires a parameter m indicating the number of models of a plurality of target models. Here, as described above, the number of models m is 2, and as will be described later, it may be expressed as the number of the internal ratio parameter vectors μ k (i) of.

[0113] Also, in step S11, the acquisition unit 11 acquires information on the prior distribution of the hidden variable z k As an example, the acquisition unit 11 acquires the prior distribution p(z k of the hidden variable zk The covariance matrix Φ of ) is obtained. In addition, the acquisition unit 11 obtains the hidden variable z k Prior distribution p(z k The covariance parameter η of ) may be obtained further.

[0114] (Step S12: Specified Processing) In step S12, the designation unit 12 selects one or more explanatory variables and one or more target variables from a plurality of features included in the target data TD.

[0115] (Step S136: Initial value determination process) Next, in step S136, the initial value determination unit 136 determines the initial value of the internal ratio parameter RP, which will be referenced in the regression coefficient calculation process S132 described later. As an example, the initial value determination unit 136 determines the initial value of the internal ratio parameter RP as a random value. By determining the initial value of the internal ratio parameter RP in this way, the regression coefficient RC can be suitably calculated in the regression coefficient calculation process S132 described later. Details of the internal ratio parameter RP have been explained, so they will be omitted here.

[0116] (Step S132: Calculation of regression coefficients) Next, in step S132, the regression coefficient calculation unit 132 calculates the internal ratio parameter (internal ratio parameter vector) RP, which is expressed by the following equation (15).

[0117]

number

[0118] The aforementioned target data TD is expressed by the following formula (16).

[0119]

number

[0120] Referring to the above, the regression coefficient RC for each of the multiple target models is expressed as equation (17) below.

[0121]

number

[0122] The following is calculated. As an example, the regression coefficient calculation unit 132 refers to the internal ratio parameter RP and the target data TD and calculates the regression coefficient RC, which is expressed as equation (19) below, using equation (18) below.

[0123]

number

[0124]

number

[0125] Here, the asterisk superscript to W indicates the updated value, and in the calculation formula, the operation symbol shown as a circle with a cross represents the Kronecker product. Also, T represents the transpose. k The hidden variable z k This represents the covariance parameter of the posterior distribution.

[0126] (Step S133: Covariance calculation process) Next, in step S133, the covariance calculation unit 133 calculates the target data TD, which is expressed by the following formula (20).

[0127]

number

[0128] The internal ratio parameter (internal ratio parameter vector) RP is expressed by the following equation (21).

[0129]

number

[0130] and the regression coefficient RC expressed by the following formula (22)

[0131]

Number

[0132] and the covariance matrix Φ of the prior distribution of the hidden variable z k Referring to the covariance parameter η of the prior distribution of the hidden variable z k and the covariance matrix {Ψ k} k} k=1 N and calculate the covariance matrix {Ψ

[0133]

Number

[0134] (where Λ k is given by the following formula (24))

[0135]

Number

[0136] to calculate the covariance matrix {Ψ k} k} k=1 N of the posterior distribution of the hidden variable z. Also, the covariance calculation unit 133 uses the following formula (25)

[0137]

Number

[0138] to calculate the covariance matrix {Ψ kWe calculate the covariance parameter η of the prior distribution. Here, N is the number of samples for each state variable as described above, and r is ~y k This is the dimension, and one example is r=1.

[0139] (Step S134: Internal division ratio parameter calculation process) Next, in step S134, the internal ratio parameter calculation unit 134 calculates the target data TD, which is expressed by the following formula (26).

[0140]

number

[0141] The regression coefficient RC is expressed by the following equation (27).

[0142]

number

[0143] And the aforementioned hidden variable z k Covariance matrix of the posterior distribution {Ψ k} k=1 N Referring to the above, the internal ratio parameter (internal ratio parameter vector) RP is expressed by the following equation (28).

[0144]

number

[0145] The following is calculated (updated). As an example, the internal division ratio parameter calculation unit 134 uses the following equation (29)

[0146]

number

[0147] by μ kThe process of calculating the internal division ratio parameter RP is performed under constraints (constraints) relating to the sum of the internal division ratio parameters RP, thereby calculating (updating) the internal division ratio parameter RP. As an example, the internal division ratio parameter calculation unit 134 uses the following equation (30)

[0148]

number

[0149] by μ k The process of calculating the internal ratio parameter (internal ratio parameter vector) RP is performed under the constraint condition expressed by equation (31) below, and the internal ratio parameter (internal ratio parameter vector) RP expressed by equation (32) below is obtained.

[0150]

number

[0151]

number

[0152] We calculate the following. Here, the first of the above constraint conditions can be expressed by explicitly specifying the index (i) related to the model, Σ i=1 m μ k (i) = 1 This can be expressed as follows. In other words, the first of the above constraint conditions indicates that the sum of the internal ratio parameter RP over the model index is 1. Also, the second of the above constraint conditions indicates that the value of the internal ratio parameter RP is 0 or greater. In this way, the internal ratio parameter calculation unit 134 can suitably calculate the internal ratio parameter RP by calculating the internal ratio parameter RP under the constraint conditions, for example, even when a normal distribution is adopted as the posterior distribution of the hidden variable.

[0153] (Step S137: Convergence determination process) Next, in step S137, the convergence determination unit 137 determines whether the series of processes in steps S132, S133, and S134 described above have converged. This can be expressed as determining whether the variational Bayes EM algorithm described above has converged, or as determining whether the calculation regarding the internal ratio parameter RP in step S134 has converged. As an example, the convergence determination unit 137 determines the variational lower bound (VLB) J obtained by the following equation (33),

[0154]

number

[0155] The system refers to the value of and determines that the series of processes performed in steps S132, S133, and S134 described above have converged if the change in the lower limit of the variation is less than or equal to a predetermined threshold. For example, in the nth convergence determination process in the repetition of the series of processes including steps S132, S133, and S134 described above, the convergence determination unit 137 compares the (n-1)th lower limit of variation with the nth lower limit of variation, and determines that the series of processes performed in steps S132, S133, and S134 described above have converged if the absolute value of the difference between them is less than or equal to a predetermined threshold.

[0156] Then, if the convergence determination unit 137 determines that "convergence has occurred," the process proceeds to output processing S135. If it determines that "convergence has not occurred," the process returns to regression coefficient calculation processing S132 and repeats the calculation of the regression coefficient RC.

[0157] (Step S135: Output processing) In step S137, if the convergence determination unit 137 determines that "convergence has occurred", then in step S135, the output unit 135 outputs the regression coefficient RC calculated by the regression coefficient calculation unit 132 in step S132, which is expressed as the following equation (34).

[0158]

number

[0159] Outputs the first linear combination LC1 and the second linear combination LC2, which are identified by [the specified method].

[0160] Thus, when the convergence determination unit 137 determines that the calculations relating to the internal division ratio parameter RP have "converged", the output unit 135 outputs the first linear sum LC1 and the second linear sum LC2, thereby outputting a suitable first linear sum LC1 and second linear sum LC2.

[0161] Furthermore, in step S135, the output unit 135 may be configured to display graphs of a first linear sum LC1 and a second linear sum LC2, which are identified by the regression coefficients RC for the two target models, in a manner that allows them to be identified from one another.

[0162] Figure 7 shows an example of a graph displayed by the output unit 135 via the input / output unit 17A in this step. In the example shown in Figure 7, the output unit 135 displays the regression coefficient W of model 1 among the regression coefficients RC calculated for each of the two target models in the regression coefficient calculation process of step S132. (1) The graph L1 of the first linear sum LC1 identified by and the regression coefficients W of Model 2. (2) The graph L2 of the second linear sum LC2, identified by [the specified method], is displayed in a way that allows them to be distinguished from each other.

[0163] Thus, according to the information processing device 1A of this exemplary embodiment, two models are used, and the regression coefficient RC can be determined by learning the internal ratio parameter RP of each model. Therefore, it is possible to generate an output result that includes the upper and lower limits of the target variable (for example, an output result that includes graphs L1 and L2 above).

[0164] (Example 2 of the processing flow in information processing device 1A) Figure 8 shows another example of the processing flow in the information processing device 1A according to this exemplary embodiment. Another example of the processing flow in the information processing device 1A will be described with reference to Figure 8.

[0165] (Steps S11 to S135) The process from step S11, in which the acquisition unit 11 acquires the target data TD, to step S135, in which the output unit 135 outputs the first linear sum LC1 and the second linear sum LC2, which are identified by the regression coefficients RC, when the convergence determination unit 137 determines that "convergence has occurred," is the same as the process described above, so its explanation is omitted.

[0166] (Step S14) In step S14, the generation unit 14 generates information about explanatory variables corresponding to one or more coefficients by referring to one or more coefficients included in at least one of the first linear combination LC1 and the second linear combination LC2.

[0167] For example, let's consider the case where the dependent variable is y, the independent variables are f={f1,f2,...}, and the first linear combination LC1 and the second linear combination LC2 are expressed as equations (35) and (36) below.

[0168]

number

[0169]

number

[0170] In this case, the generation unit 14 generates information about the explanatory variable f corresponding to at least one of the coefficients a={a1, a2, ...} and the coefficients c={c1, c2, ...} as explanatory information EI.

[0171] As an example, we will explain using a case where one or more dependent variables y include one or more indicators related to the work, and one or more independent variables f={f1,f2,...} include features related to the workers performing the work.

[0172] For example, consider a scenario where explanatory variable f1 is the increase in working time when worker A1, corresponding to coefficient a1, is a worker performing the task, and explanatory variable f2 is the increase in working time when worker A2, corresponding to coefficient a2, is a worker performing the task. In other words, consider a scenario where, if worker A1 is a worker, working time increases by f1 (if coefficient a1 is negative, working time decreases by f1), and if worker A2 is a worker, working time increases by f2 (if coefficient a2 is negative, working time decreases by f1).

[0173] In this case, coefficient a1 and coefficient a2 These represent the skill levels of worker A1 and worker A2 in their respective tasks. Therefore, if the coefficient a is negative and its absolute value is greater than a predetermined value, the generation unit 14 generates explanatory information EI indicating that the skill level of the worker corresponding to coefficient a is 3 (high skill level). Similarly, if the absolute value of coefficient a is less than or equal to a predetermined value, the generation unit 14 generates explanatory information EI indicating that the skill level of the worker corresponding to coefficient a is 2 (medium skill level). Furthermore, if the coefficient a is positive and its absolute value is greater than a predetermined value, the generation unit 14 generates explanatory information EI indicating that the skill level of the worker corresponding to coefficient a is 1 (low skill level).

[0174] Figure 9 shows an example of the graph displayed by the output unit 135 via the input / output unit 17A in step S135, and the explanatory information EI displayed by the generation unit 14 via the input / output unit 17A in step S14. In the example shown in Figure 9, similar to Figure 7 described above, the output unit 135 displays the regression coefficients RC, specifically the regression coefficients W of Model 1. (1) The graph L1 of the first linear sum LC1 identified by and the regression coefficients W of Model 2. (2)The graph L2 of the second linear sum LC2 specified by is displayed so as to be distinguishable from each other. Further, in the example shown in FIG. 9, the generation unit 14 displays, as the explanatory information EI, a table indicating that the proficiency level of the worker with the worker ID "001" is "1", the proficiency level of the worker with the worker ID "002" is "3", and the proficiency level of the worker with the worker ID "003" is "2".

[0175] Thus, according to the information processing apparatus 1A according to this exemplary embodiment, it is possible to generate the explanatory information EI in which information difficult to quantify such as the proficiency level is quantified.

[0176] (Example of processing in the optimization apparatus 60) The optimization unit 63 of the optimization apparatus 60 acquires the first linear sum LC1 and the second linear sum LC2 output from the information processing apparatus 1A. Then, the optimization unit 63 executes an optimization process by referring to at least a part of the target data TD.

[0177] For example, assume that the target data TD is log data in which worker A = {A1, A2,...} performed work X, the objective variable y is the work time, the explanatory variables f = {f1, f2,...} are the work times that increase when the corresponding worker A = {A1, A2,...} performs the work, the first linear sum LC1 and the second linear sum LC2 are expressed by the following equations (37) and (38), and the first linear sum LC1 defines the upper limit of the work time of work X, and the second linear sum LC2 defines the lower limit of the work time of work X.

[0178]

Equation

[0179]

Equation

[0180] In this case, the optimization unit 63 optimizes the shifts of worker A = {A1, A2, ...}. The optimization unit 63 may optimize the shifts of worker A = {A1, A2, ...} without constraints. Furthermore, if the optimization process results in a combinatorial explosion (no solution is found), the optimization device 60 may derive constraints defined using at least one of the first linear combination LC1 and the second linear combination LC2, and execute the optimization process under those constraints.

[0181] Thus, the optimization device 60 according to this exemplary embodiment can solve optimization problems such as optimizing work shifts.

[0182] (Effects of Information Processing Device 1A) As described above, the information processing device 1A specifies one or more explanatory variables and one or more target variables from multiple features included in the target data TD, and derives a first linear combination LC1 which defines the upper limit of the target variable, which is the first linear combination LC1 of one or more explanatory variables, and a second linear combination LC2 which defines the lower limit of the target variable, which is the second linear combination LC2 of one or more explanatory variables.

[0183] With this configuration, even for target data TD containing multiple features, the information processing device 1A can specify explanatory variables and a target variable from the multiple features and derive a linear sum of the explanatory variables that defines the upper and lower limits of the target variable. Therefore, the information processing device 1A can set appropriate upper and lower limits for data containing multiple features.

[0184] [Third Exemplary Embodiment] A third exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same function as those described in the above-described exemplary embodiments are denoted by the same reference numerals, and their descriptions are omitted as appropriate. The scope of application of each technology adopted in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology adopted in this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical hindrance occurs. Furthermore, each technology shown in the drawings referenced to describe this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical hindrance occurs.

[0185] (Overview of Information Processing Device 1B) The information processing device 1B includes the configuration of the optimization device 60 in addition to the configuration of the information processing device 1A described above.

[0186] (Configuration of Information Processing Unit 1B) The configuration of the information processing device 1B will be described with reference to Figure 10. Figure 10 is a block diagram showing the configuration of the information processing device 1B. As shown in Figure 10, the information processing device 1B includes a control unit 10B, a storage unit 15A, a communication unit 16A, and an input / output unit 17A. The storage unit 15A, the communication unit 16A, and the input / output unit 17A are as described above.

[0187] The control unit 10B controls each component of the information processing device 1B. The control unit 10B also includes an acquisition unit 11, a designation unit 12, a derivation unit 13, and an optimization unit 63, as shown in Figure 10. In this exemplary embodiment, the acquisition unit 11, designation unit 12, derivation unit 13, and optimization unit 63 implement the acquisition means, designation means, derivation means, and optimization means, respectively. The acquisition unit 11, designation unit 12, and derivation unit 13 are described above.

[0188] The optimization unit 63, similar to the optimization unit 63 in the optimization device 60 in the exemplary embodiment described above, performs an optimization process that references at least a portion of the target data TD under constraints defined using at least one of the first linear sum LC1 and the second linear sum LC2.

[0189] (Effects of Information Processing Device 1B) As described above, the information processing device 1B, like the information processing device 1A, specifies one or more explanatory variables and one or more target variables from multiple features contained in the target data TD, and derives a first linear combination LC1 which defines the upper limit of the target variable, consisting of one or more explanatory variables, and a second linear combination LC2 which defines the lower limit of the target variable, consisting of one or more explanatory variables. Therefore, the information processing device 1B, like the information processing device 1A, can set appropriate upper and lower limits for data containing multiple features.

[0190] Furthermore, the information processing device 1B is equipped with an optimization unit 63. Therefore, the information processing device 1B can solve optimization problems, such as optimizing shifts in business operations.

[0191] [Examples of implementation using software] Some or all of the functions of the information processing devices 1, 1A, 1B, the first information processing device 1, the second information processing device 2, and the optimization device 60 (hereinafter also referred to as "each of the above devices") may be implemented by hardware such as integrated circuits (IC chips) or by software.

[0192] In the latter case, each of the above devices is implemented, for example, by a computer that executes instructions for a program, which is software that realizes each function. An example of such a computer (hereinafter referred to as computer C) is shown in Figure 11. Figure 11 is a block diagram showing the hardware configuration of computer C, which functions as each of the above devices.

[0193] Computer C comprises at least one processor C1 and at least one memory C2. Memory C2 stores a program P that causes computer C to operate as each of the above-mentioned devices. In computer C, processor C1 reads program P from memory C2 and executes it, thereby realizing each of the above-mentioned devices.

[0194] For processor C1, for example, a CPU (Central Processing Unit), GPU (Graphic Processing Unit), DSP (Digital Signal Processor), MPU (Micro Processing Unit), FPU (Floating Point Number Processing Unit), PPU (Physics Processing Unit), TPU (Tensor Processing Unit), quantum processor, microcontroller, or a combination thereof can be used. For memory C2, for example, flash memory, HDD (Hard Disk Drive), SSD (Solid State Drive), or a combination thereof can be used.

[0195] Computer C may also be equipped with RAM (Random Access Memory) for loading program P at runtime and for temporarily storing various data. Furthermore, computer C may be equipped with communication interfaces for sending and receiving data with other devices. Additionally, computer C may be equipped with input / output interfaces for connecting input / output devices such as keyboards, mice, displays, and printers.

[0196] Also, the program P can be recorded on a non-transitory tangible recording medium M readable by the computer C. As such a recording medium M, for example, a tape, a disk, a card, a semiconductor memory, or a programmable logic circuit can be used. The computer C can acquire the program P via such a recording medium M. Also, the program P can be transmitted via a transmission medium. As such a transmission medium, for example, a communication network or a broadcast wave can be used. The computer C can also acquire the program P via such a transmission medium.

[0197] Also, each of the above functions of each of the above devices may be realized by a single processor provided in a single computer, or may be realized by a plurality of processors provided in a single computer cooperating with each other, or may be realized by a plurality of processors provided in each of a plurality of computers cooperating with each other. Also, the program for causing each of the above devices to realize each of the above functions may be stored in a single memory provided in a single computer, or may be distributed and stored in a plurality of memories provided in a single computer, or may be distributed and stored in a plurality of memories provided in each of a plurality of computers.

[0198] 〔Supplementary Note A〕 The present disclosure includes the technologies described in the following supplementary notes. However, the present invention is not limited to the technologies described in the following supplementary notes, and various modifications are possible within the scope indicated in the claims.

[0199] (Supplementary Note A1) An acquisition unit that acquires target data including a plurality of feature amounts, A specification unit that specifies one or more explanatory variables and one or more objective variables from the plurality of feature amounts included in the target data, A first linear sum that defines an upper limit of the objective variable, which is a first linear sum of the one or more explanatory variables, A second linear sum that defines a lower limit of the objective variable, which is a second linear sum of the one or more explanatory variables A means for deriving the and An information processing device equipped with the following features.

[0200] (Appendix A2) The system further comprises generating means for generating information about explanatory variables corresponding to one or more coefficients, by referencing one or more coefficients included in at least one of the first linear combination and the second linear combination. The information processing device described in Appendix A1.

[0201] (Note A3) The aforementioned one or more dependent variables include one or more indicators related to the business, The one or more explanatory variables mentioned above include features related to the workers performing the tasks, The generation means generates information regarding the worker's proficiency as information regarding the explanatory variables. The information processing device described in Appendix A2.

[0202] (Note A4) The system further includes optimization means that perform an optimization process that references at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination. An information processing device as described in any one of the appendices A1 to A3.

[0203] (Note A5) The derivation means is Multiple regression models, each associated with a multiple ratio parameter defined by a hidden variable, are trained by referring to at least a portion of the target data. The first linear combination and the second linear combination are derived using the plurality of regression models. An information processing device as described in any one of the appendices A1 to A4.

[0204] (Note A6) The acquisition means further acquires information regarding the prior distribution of the hidden variable, The derivation means is, A regression coefficient calculation means calculates the regression coefficient for each of the plurality of regression models by referring to the ratio parameter that defines the internal ratio of the plurality of regression models and the target data, A covariance calculation means that calculates the covariance parameter of the prior distribution of the hidden variable and the covariance matrix of the posterior distribution of the hidden variable by referring to the target data, the ratio parameter, the regression coefficient, and information regarding the prior distribution of the hidden variable, A ratio parameter calculation means that calculates the ratio parameter by referring to the target data, the regression coefficient, and the covariance matrix of the posterior distribution of the hidden variable, An information processing device described in Appendix A5, which is equipped with the following features.

[0205] (Note A7) An information processing system including a first information processing device and a second information processing device, The first information processing device is A means for acquiring target data that includes multiple features, A selection means for selecting one or more explanatory variables and one or more target variables from the multiple features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables and A means for deriving the and Equipped with, The first second information processing device is The system includes an optimization means that performs an optimization process that references at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination. Information processing system.

[0206] (Note A8) A means for acquiring target data that includes multiple features, A selection means for selecting one or more explanatory variables and one or more target variables from the multiple features included in the target data, A generating means that generates information about the explanatory variables corresponding to the one or more coefficients by referring to one or more coefficients included in at least one of a first linear combination defining the upper limit of the dependent variable, which is a first linear combination of one or more explanatory variables, and a second linear combination defining the lower limit of the dependent variable, which is a second linear combination of one or more explanatory variables. An information processing device equipped with the following features.

[0207] (Note A9) The aforementioned one or more dependent variables include one or more indicators related to the business, The one or more explanatory variables mentioned above include features related to the workers performing the tasks, The generation means generates information regarding the worker's proficiency as information regarding the explanatory variables. The information processing device described in Appendix A8.

[0208] [Additional Notes B] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0209] (Note B1) At least one processor performs an acquisition process to obtain target data containing multiple features, The at least one processor performs a designation process to specify one or more explanatory variables and one or more target variables from the plurality of features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables and The derivation process performed by the at least one processor and An information processing method that includes this.

[0210] (Note B2) The at least one processor further includes a generation process that references one or more coefficients included in at least one of the first linear combination and the second linear combination to generate information about explanatory variables corresponding to said one or more coefficients. The information processing method described in Appendix B1.

[0211] (Note B3) The aforementioned one or more dependent variables include one or more indicators related to the business, The one or more explanatory variables mentioned above include features related to the workers performing the tasks, In the generation process, the at least one processor generates information regarding the worker's proficiency as information regarding the explanatory variables. The information processing method described in Appendix B2.

[0212] (Note B4) The at least one processor further includes an optimization process that performs an optimization process that references at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination. The information processing method described in any one of the appendices B1 to B3.

[0213] (Note B5) In the derivation process, at least one processor, Multiple regression models, each associated with a multiple ratio parameter defined by a hidden variable, are trained by referring to at least a portion of the target data. The first linear combination and the second linear combination are derived using the plurality of regression models. The information processing method described in any one of the appendices B1 to B4.

[0214] (Note B6) In the acquisition process, the at least one processor further acquires information regarding the prior distribution of the hidden variables, The aforementioned derivation process is, A regression coefficient calculation process that calculates the regression coefficient for each of the plurality of regression models by referring to the ratio parameter that defines the internal ratio of the plurality of regression models and the target data, A covariance calculation process that calculates the covariance parameter of the prior distribution of the hidden variable and the covariance matrix of the posterior distribution of the hidden variable by referring to the target data, the ratio parameter, the regression coefficient, and information regarding the prior distribution of the hidden variable, A ratio parameter calculation process that calculates the ratio parameter by referring to the target data, the regression coefficients, and the covariance matrix of the posterior distribution of the hidden variables, The information processing method described in Appendix B5, which includes the information processing method described therein.

[0215] (Note B8) At least one processor performs an acquisition process to obtain target data containing multiple features, The at least one processor performs a selection process to select one or more explanatory variables and one or more target variables from the plurality of features included in the target data, The at least one processor generates information about the explanatory variables corresponding to the one or more coefficients by referencing one or more coefficients included in at least one of a first linear combination defining the upper limit of the objective variable, which is a first linear combination of one or more explanatory variables, and a second linear combination defining the lower limit of the objective variable, which is a second linear combination of one or more explanatory variables. An information processing method that includes this.

[0216] (Note B9) The aforementioned one or more dependent variables include one or more indicators related to the business, The one or more explanatory variables mentioned above include features related to the workers performing the tasks, In the generation process, the at least one processor generates information regarding the worker's proficiency as information regarding the explanatory variables. The information processing method described in Appendix B8.

[0217] [Additional Note C] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0218] (Note C1) A program that makes a computer function as an information processing device. The aforementioned computer, A means for acquiring target data that includes multiple features, A specifying means for specifying one or more explanatory variables and one or more target variables from the multiple features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables and A means for deriving the and An information processing program that functions as such.

[0219] (Note C2) The aforementioned computer, The system is further configured to function as a generating means for generating information about explanatory variables corresponding to one or more coefficients, by referencing one or more coefficients included in at least one of the first linear combination and the second linear combination. The information processing program described in Appendix C1.

[0220] (Note C3) The aforementioned one or more dependent variables include one or more indicators related to the business, The one or more explanatory variables mentioned above include features related to the workers performing the tasks, The generation means generates information regarding the worker's proficiency as information regarding the explanatory variables. The information processing program described in Appendix C2.

[0221] (Note C4) The aforementioned computer, It is further configured to function as an optimization means that performs an optimization process that references at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination. An information processing program described in any one of the appendices C1 to C3.

[0222] (Note C5) The derivation means is Multiple regression models, each associated with a multiple ratio parameter defined by a hidden variable, are trained by referring to at least a portion of the target data. The first linear combination and the second linear combination are derived using the plurality of regression models. An information processing program described in any one of the appendices C1 to C4.

[0223] (Appendix C6) The aforementioned computer, The acquisition means further acquires information regarding the prior distribution of the hidden variable, The derivation means is, A regression coefficient calculation means calculates the regression coefficient for each of the plurality of regression models by referring to the ratio parameter that defines the internal ratio of the plurality of regression models and the target data, A covariance calculation means that calculates the covariance parameter of the prior distribution of the hidden variable and the covariance matrix of the posterior distribution of the hidden variable by referring to the target data, the ratio parameter, the regression coefficient, and information regarding the prior distribution of the hidden variable, A ratio parameter calculation process that calculates the ratio parameter by referring to the target data, the regression coefficients, and the covariance matrix of the posterior distribution of the hidden variables, The information processing program described in Appendix C5 is used to perform this function.

[0224] (Note C7) A program that makes a computer function as an information processing system, The aforementioned computer, A means for acquiring target data that includes multiple features, A selection means for selecting one or more explanatory variables and one or more target variables from the multiple features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables and A means for deriving the and An optimization means that performs an optimization process that references at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination; To make it function as Information processing program.

[0225] (Note C8) A program that makes a computer function as an information processing device. The aforementioned computer, A means for acquiring target data that includes multiple features, A selection means for selecting one or more explanatory variables and one or more target variables from the multiple features included in the target data, A generating means that generates information about the explanatory variables corresponding to the one or more coefficients by referring to one or more coefficients included in at least one of a first linear combination defining the upper limit of the dependent variable, which is a first linear combination of one or more explanatory variables, and a second linear combination defining the lower limit of the dependent variable, which is a second linear combination of one or more explanatory variables. An information processing program that functions as such.

[0226] (Note C9) The aforementioned one or more dependent variables include one or more indicators related to the business, The one or more explanatory variables mentioned above include features related to the workers performing the tasks, The generation means generates information regarding the worker's proficiency as information regarding the explanatory variables. The information processing program described in Appendix C8.

[0227] [Additional Note D] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0228] (Note D1) It comprises at least one processor, and the at least one processor is The acquisition process involves obtaining target data that includes multiple features, A specification process that specifies one or more explanatory variables and one or more target variables from the multiple features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables and The derivation process that derives and An information processing device that performs the following actions.

[0229] The information processing device may also include memory. Furthermore, the memory may store a program that causes at least one processor to execute each of the aforementioned processes.

[0230] (Note D2) The aforementioned at least one processor, Further generation processes are performed to generate information about explanatory variables corresponding to one or more coefficients, by referencing one or more coefficients included in at least one of the first linear combination and the second linear combination. The information processing device described in Appendix D1.

[0231] (Note D3) The aforementioned one or more dependent variables include one or more indicators related to the business, The one or more explanatory variables mentioned above include features related to the workers performing the tasks, In the generation process, the at least one processor generates information regarding the worker's proficiency as information regarding the explanatory variables. The information processing device described in Appendix D2.

[0232] (Note D4) The at least one processor further performs an optimization process that references at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination. An information processing device as described in any one of the appendices D1 to D3.

[0233] (Note D5) In the derivation process, at least one processor, Multiple regression models, each associated with a multiple ratio parameter defined by a hidden variable, are trained by referring to at least a portion of the target data. The first linear combination and the second linear combination are derived using the plurality of regression models. An information processing device as described in any one of the appendices D1 to D4.

[0234] (Note D6) In the acquisition process, the at least one processor further acquires information regarding the prior distribution of the hidden variables, In the derivation process, at least one processor, A regression coefficient calculation process that calculates the regression coefficient for each of the plurality of regression models by referring to the ratio parameter that defines the internal ratio of the plurality of regression models and the target data, A covariance calculation process that calculates the covariance parameter of the prior distribution of the hidden variable and the covariance matrix of the posterior distribution of the hidden variable by referring to the target data, the ratio parameter, the regression coefficient, and information regarding the prior distribution of the hidden variable, A ratio parameter calculation process that calculates the ratio parameter by referring to the target data, the regression coefficients, and the covariance matrix of the posterior distribution of the hidden variables, The information processing device described in Appendix D5 that performs the execution.

[0235] (Note D7) An information processing system including a first information processing device and a second information processing device, The first information processing device comprises at least one processor, The acquisition process involves obtaining target data that includes multiple features, A selection process that selects one or more explanatory variables and one or more target variables from the multiple features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables and The derivation process that derives and Execute, The second information processing device comprises at least one processor, An optimization process is performed that references at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination. Information processing system.

[0236] (Note D8) It comprises at least one processor, and the at least one processor is The acquisition process involves obtaining target data that includes multiple features, A selection process that selects one or more explanatory variables and one or more target variables from the multiple features included in the target data, A generation process that generates information about the explanatory variables corresponding to the one or more coefficients by referring to one or more coefficients included in at least one of the following: a first linear combination that defines the upper limit of the dependent variable, which is a first linear combination of one or more explanatory variables, and a second linear combination that defines the lower limit of the dependent variable, which is a second linear combination of one or more explanatory variables. An information processing device that performs the following actions.

[0237] The information processing device may also include memory. Furthermore, the memory may store a program that causes at least one processor to execute each of the aforementioned processes.

[0238] (Note D9) The aforementioned one or more dependent variables include one or more indicators related to the business, The one or more explanatory variables mentioned above include features related to the workers performing the tasks, In the generation process, the at least one processor generates information regarding the worker's proficiency as information regarding the explanatory variables. The information processing device described in Appendix D8.

[0239] [Additional Note E] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0240] (Note E1) A program that makes a computer function as an information processing device. To the aforementioned computer, The acquisition process involves obtaining target data that includes multiple features, A specification process that specifies one or more explanatory variables and one or more target variables from the multiple features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables and The derivation process that derives and A non-temporary recording medium that stores an information processing program that executes that program.

[0241] (Note E2) A program that makes a computer function as an information processing device. To the aforementioned computer, The acquisition process involves obtaining target data that includes multiple features, A selection process that selects one or more explanatory variables and one or more target variables from the multiple features included in the target data, A generation process that generates information about the explanatory variables corresponding to the one or more coefficients by referring to one or more coefficients included in at least one of the following: a first linear combination that defines the upper limit of the dependent variable, which is a first linear combination of one or more explanatory variables, and a second linear combination that defines the lower limit of the dependent variable, which is a second linear combination of one or more explanatory variables. A non-temporary recording medium that stores an information processing program that executes that program. [Explanation of Symbols]

[0242] 1, 1A, 1B Information Processing Devices 1. First Information Processing Device 2. Second Information Processing Device 11 Acquisition Department 12 Specified part 13 Derivation part 14 Generation part 21, 63 Optimization section 60 Optimization device 100, 100A Information Processing System 132 Regression Coefficient Calculation Unit 133 Covariance calculation part 134 Internal Ratio Parameter Calculation Unit 135 Output section 136 Initial Value Determination Unit 137 Convergence Determination Unit TD Target Data RP internal ratio parameter RC regression coefficient DI distribution information LR learning results LC1 First linear combination LC2 Second linear combination EI Explanation Information

Claims

1. A means for acquiring target data that includes multiple features, A specifying means for specifying one or more explanatory variables and one or more target variables from the multiple features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables A means for deriving the and An information processing device equipped with the following features.

2. The system further comprises generating means for generating information about explanatory variables corresponding to one or more coefficients, by referencing one or more coefficients included in at least one of the first linear combination and the second linear combination. The information processing apparatus according to claim 1.

3. The aforementioned one or more dependent variables include one or more indicators related to the business, The one or more explanatory variables mentioned above include features related to the workers performing the tasks, The generation means generates information regarding the worker's proficiency as information regarding the explanatory variables. The information processing apparatus according to claim 2.

4. The system further includes optimization means that perform an optimization process that references at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination. An information processing apparatus according to any one of claims 1 to 3.

5. The derivation means is Multiple regression models, each associated with a multiple ratio parameter defined by a hidden variable, are trained by referring to at least a portion of the target data. The first linear combination and the second linear combination are derived using the plurality of regression models. An information processing apparatus according to any one of claims 1 to 3.

6. The acquisition means further acquires information regarding the prior distribution of the hidden variable, The derivation means is, A regression coefficient calculation means calculates the regression coefficient for each of the plurality of regression models by referring to the ratio parameter that defines the internal ratio of the plurality of regression models and the target data, A covariance calculation means that calculates the covariance parameter of the prior distribution of the hidden variable and the covariance matrix of the posterior distribution of the hidden variable by referring to the target data, the ratio parameter, the regression coefficient, and information regarding the prior distribution of the hidden variable, A ratio parameter calculation means that calculates the ratio parameter by referring to the target data, the regression coefficient, and the covariance matrix of the posterior distribution of the hidden variable, The information processing apparatus according to claim 5, comprising:

7. An information processing system including a first information processing device and a second information processing device, The first information processing device is A means for acquiring target data that includes multiple features, A selection means for selecting one or more explanatory variables and one or more target variables from the multiple features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables A means for deriving the and Equipped with, The aforementioned second information processing device is The system includes an optimization means that performs an optimization process that references at least a portion of the target data, under constraints defined using at least one of the first linear combination and the second linear combination. Information processing system.

8. At least one processor performs an acquisition process to obtain target data containing multiple features, The at least one processor performs a designation process to specify one or more explanatory variables and one or more target variables from the plurality of features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables The derivation process performed by the at least one processor and An information processing method that includes this.

9. A program that makes a computer function as an information processing device. The aforementioned computer, A means for acquiring target data that includes multiple features, A specifying means for specifying one or more explanatory variables and one or more target variables from the multiple features included in the target data, A first linear combination that defines the upper limit of the dependent variable, comprising the first linear combination of the one or more explanatory variables, A second linear combination that defines the lower limit of the dependent variable, wherein the second linear combination of the one or more explanatory variables A means for deriving the and An information processing program that functions as such.

Images