Learning device, presentation device, learning method and program

Abstract

A training device 1X mainly comprises a solution generation means 15X and a training means 16X. The solution generation means 15X generates a solution of an optimization problem. The training means 16X generates multiple sets of solutions. In addition, the training means 16X trains a model for determining a set of solutions to be outputted, on the basis of a set selected from the multiple sets by external input.

Description

Technical Field

[0001] The present disclosure relates to the technical field of a learning device, a presentation device, a learning method, and a storage medium that perform processing related to optimization problems.

Background Art

[0002] Systems for calculating solutions to optimization problems have been conventionally known. For example, Patent Document 1 discloses a technique for evaluating optimization results based on a plurality of indicators. In addition, Patent Document 2 discloses an optimization system that can determine (match) combinations such that transaction conditions such as transaction volume and transaction price desired by a seller and a buyer of goods to be traded match each other, and then perform rematching by changing some transaction conditions. Further, Patent Document 3 discloses a system that optimizes business scheduling in consideration of a plurality of indicators.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

Problem to be Solved by the Invention

[0004] When a solution to an optimization problem is obtained through optimization, a solution that satisfies the formulated constraints can be obtained. On the other hand, a user may want to compare and examine a plurality of solutions to determine a solution to be finally adopted.

[0005] One of the purposes of this disclosure is, in light of the above-mentioned problems, to provide a learning device, a learning method, a storage medium, and a presentation device that presents results based on the learning outcome, which are used to learn how to present multiple solutions with variations preferred by the user. [Means for solving the problem]

[0006] One aspect of a learning device is: Solution generation means for generating solutions to optimization problems, Multiple sets of the aforementioned solution are generated, and each of the multiple sets Regarding the set, the multiple solutions included in the set, and the values ​​of the items that are indicators or variables to be obtained as solutions used in the objective function or constraints of the optimization problem for each of the multiple solutions, The system presents the multiple sets of input from the user. Inside A learning means for learning a model that determines the set of solutions to output based on the set selected from, It is a learning device that has [a certain feature].

[0007] One aspect of the learning method is: Computers To generate a solution to an optimization problem, Multiple sets of the aforementioned solution are generated, and each of the multiple sets Regarding the set, the multiple solutions included in the set, and the values ​​of the items that are indicators or variables to be obtained as solutions used in the objective function or constraints of the optimization problem for each of the multiple solutions, The system presents the multiple sets of input from the user. Inside A model is trained to determine the set of solutions to output based on the set selected from the above. This is a learning method.

[0008] One aspect of the program is: To generate a solution to an optimization problem, Multiple sets of the aforementioned solution are generated, and each of the multiple sets Regarding the set, the multiple solutions included in the set, and the values ​​of the items that are indicators or variables to be obtained as solutions used in the objective function or constraints of the optimization problem for each of the multiple solutions, The system presents the multiple sets of input from the user. Inside This program causes a computer to perform a process of learning a model that determines the set of solutions to be output based on the set selected from the above. [Effects of the Invention]

[0009] It can learn to present multiple solutions with variations that users prefer. [Brief explanation of the drawing]

[0010] [Figure 1] Shows the configuration of the optimization system according to the first embodiment. [Figure 2] Shows the hardware configuration of the information processing apparatus. [Figure 3] It is an example of a functional block diagram according to the first embodiment. [Figure 4] It is an example of a flowchart of the learning phase according to the first embodiment. [Figure 5] It is an example of a flowchart of the presentation phase according to the first embodiment. [Figure 6] It is a display example of a solution selection screen. [Figure 7] It is a display example of a solution presentation screen. [Figure 8] It is a display example of a ranking screen. [Figure 9] Shows the configuration of the optimization system according to the second embodiment. [Figure 10] It is a functional block diagram of the learning apparatus according to the third embodiment. [Figure 11] It is an example of a flowchart showing the processing procedure of the learning apparatus according to the third embodiment. DESCRIPTION OF EMBODIMENTS

[0011] Hereinafter, embodiments of a learning apparatus, a presentation apparatus, a learning method, and a storage medium will be described with reference to the drawings.

[0012] <First Embodiment> (1) System Configuration Figure 1 shows the configuration of an optimization system 100 according to the first embodiment. The optimization system 100 mainly includes an information processing apparatus 1, an input apparatus 2, a display apparatus 3, and a storage apparatus 4.

[0013] The information processing device 1 performs optimization processing on a specified optimization problem. In this embodiment, the information processing device 1 calculates multiple solutions to the specified optimization problem and presents the multiple solutions (which may also be sequences of solution transitions; the same applies hereinafter). In this case, when presenting the multiple solutions, the information processing device 1 learns to present multiple solutions that have variations that are important to the user using the optimization system 100. Hereafter, the stage (phase) in which learning is performed to present multiple solutions that have variations important to the user will be called the "learning phase," and the phase in which multiple solutions for the optimization problem specified by the user are presented using the parameters obtained through learning will also be called the "presentation phase." Note that the learning phase is not a mandatory process, and the user can choose whether or not to perform it. The information processing device 1 is an example of a "learning device" and a "presentation device."

[0014] The optimization problem may be, for example, a problem of determining the combination of sellers and buyers of goods to be traded (and the transportation schedule for those goods), a problem of determining employee work shifts, or any other arbitrary combinatorial optimization problem. The goods to be traded as described above may be fuels such as LNG, steel, machinery, electronics, textiles, chemical products, medical-related goods, food, or any other items.

[0015] Furthermore, the information processing device 1 communicates data with the input device 2, the display device 3, and the storage device 4 via a communication network or through direct wireless or wired communication.

[0016] Input device 2 is an interface that accepts user input, which is an external input, and includes, for example, a touch panel, buttons, a keyboard, or a voice input device. Input device 2 supplies input information "S1" generated based on the user's input to information processing device 1.

[0017] The display device 3 is, for example, a display, a projector, etc., and performs a predetermined display based on the display information "S2" supplied from the information processing device 1.

[0018] The storage device 4 is a memory that stores various information necessary for the optimization process. For example, the storage device 4 stores condition information 40 and parameter information 41.

[0019] Condition information 40 is information about the conditions of the optimization problem that the information processing device 1 should solve (including information about the problem setting). For example, if the optimization problem that the information processing device 1 should solve is set as the problem of determining the combination of sellers and buyers of goods to be traded, then condition information 40 will include information about the sellers of the goods to be traded (including each seller's desired conditions regarding the delivery place, delivery period, transaction quantity, and price, etc.), information about the buyers of the goods to be traded (including each buyer's desired conditions regarding the delivery place, delivery period, transaction quantity, and price, etc.). Note that condition information 40 may store condition information for the optimization problem targeted in the learning phase and condition information for the optimization problem targeted in the presentation phase, respectively.

[0020] Parameter information 41 is information about the parameters of a model (also called a "presented solution decision model") that determines which of the multiple solutions (also called "presented solutions") to present to the user. The presented solution decision model is a learning model that, for example, is trained to output information indicating which set of multiple solutions should be presented when multiple sets of candidate solutions are input. The presented solution decision model may have the architecture of any machine learning model, such as a neural network or a support vector machine. For example, if the presented solution decision model has the architecture of a neural network such as a convolutional neural network, the parameter information 41 includes information such as the layer structure, the neuron structure of each layer, the number and size of filters in each layer, and the weights of each element of each filter. Before the learning phase is performed, information indicating predetermined initial parameters may be stored in the storage device 4 as parameter information 41.

[0021] The storage device 4 may be an external storage device such as a hard disk connected to or built into the information processing device 1, or it may be a storage medium such as flash memory. Furthermore, the storage device 4 may be a server device that communicates data with the information processing device 1. In this case, the storage device 4 may consist of multiple server devices.

[0022] The configuration of the optimization system 100 shown in Figure 1 is an example, and various modifications may be made to this configuration. For example, the input device 2 and the display device 3 may be configured as a single unit. In this case, the input device 2 and the display device 3 may be configured as a tablet terminal integrated with the information processing device 1. Furthermore, the information processing device 1 may be composed of multiple devices. In this case, the multiple devices constituting the information processing device 1 exchange information among themselves that is necessary to execute pre-assigned processes.

[0023] (2) Hardware configuration of information processing equipment Figure 2 shows the hardware configuration of the information processing device 1. The information processing device 1 includes a processor 11, memory 12, and interface 13 as hardware components. The processor 11, memory 12, and interface 13 are connected via a data bus 19.

[0024] The processor 11 executes predetermined processes by running programs stored in memory 12. The processor 11 is a processor such as a CPU (Central Processing Unit), GPU (Graphics Processing Unit), or TPU (Tensor Processing Unit). The processor 11 may be composed of multiple processors. The processor 11 is an example of a computer.

[0025] Memory 12 is composed of various volatile and non-volatile memories, such as RAM (Random Access Memory) and ROM (Read Only Memory). Memory 12 also stores programs for the information processing device 1 to execute various processes. Furthermore, Memory 12 is used as working memory to temporarily store information obtained from the storage device 4. Memory 12 may also function as the storage device 4. Similarly, the storage device 4 may function as the memory 12 of the information processing device 1. Note that the programs executed by the information processing device 1 may be stored in storage media other than memory 12.

[0026] Interface 13 is an interface for electrically connecting the information processing device 1 with other devices. These interfaces may be wireless interfaces such as network adapters for wirelessly transmitting and receiving data with other devices, or they may be hardware interfaces for connecting with other devices via cables, etc.

[0027] The hardware configuration of the information processing device 1 is not limited to the configuration shown in Figure 2. For example, the information processing device 1 may include at least one of the input device 2 or the display device 3. Furthermore, the information processing device 1 may be connected to or have a built-in sound output device such as a speaker.

[0028] (3) Functional Blocks Figure 3 shows an example of the functional blocks of the processor 11 in the first embodiment. Functionally, the processor 11 includes a solution generation unit 15, a learning unit 16, a solution presentation determination unit 17, and a UI (User Interface) control unit 18. In Figure 3, blocks that exchange data are connected by solid lines, but the combination of blocks that exchange data is not limited to Figure 3. The same applies to the diagrams of other functional blocks described later.

[0029] The solution generation unit 15 determines the solution to the optimization problem. In this case, the solution generation unit 15 calculates multiple solutions to the optimization problem based on the condition information 40 of the optimization problem stored in the storage device 4. In this case, the solution generation unit 15 calculates a predetermined number of solutions, for example, more than the number of solutions to be presented. Alternatively, in order to obtain a predetermined number of solutions, the solution generation unit 15 may relax the constraints stored in the condition information 40 and find an optimal solution for each relaxed constraint. The degree of constraint relaxation in this case may be determined based on user input.

[0030] The solution generation unit 15 may determine the solution to the optimization problem based on any optimization method (optimization solver). For example, when the solution generation unit 15 solves the problem of determining the combination of sellers and buyers of goods to be traded, it considers this problem as a single combinatorial optimization problem and formulates it as an integer programming problem. The solution generation unit 15 then finds the solution by performing processing equivalent to that of a general application program (e.g., IBM ILOG CPLEX, Gurobi Optimizer, SCIP) on the formulated integer programming problem. A method for formulating an integer programming problem to determine the seller, buyer, vessel to be used, and the duration of the vessel's voyage is disclosed, for example, in Patent Document 2.

[0031] The learning unit 16 updates parameter information 41 during the learning phase. In this case, the learning unit 16 generates multiple sets of multiple solutions (i.e., generates multiple candidate solutions) based on the calculation results of the solution generation unit 15, and presents these to the user in a selectable format. In this case, the learning unit 16 generates multiple sets of multiple solutions, for example, by randomly sampling from the solutions calculated by the solution generation unit 15. The learning unit 16 then considers the set of multiple solutions selected by the user input from the multiple sets of multiple solutions presented as the correct solution (i.e., a solution with variations preferred by the user). As a result, the learning unit 16 generates training data for a solution decision model, using the multiple sets of multiple solutions presented to the user as input data and the set of multiple solutions selected by the user input as the correct answer data. In training the solution decision model using the above-mentioned training data, the learning unit 16 determines the parameters of the solution decision model so as to minimize the error (loss) between the data output by the solution decision model when input data is received and the correct answer data corresponding to the input data. The algorithm used to determine the parameters mentioned above in order to minimize loss may be any learning algorithm used in machine learning, such as gradient descent or backpropagation.

[0032] The solution presentation unit 17 determines N (where N is an integer greater than or equal to 2) solutions to be presented to the user during the presentation phase. In this case, the solution presentation unit 17 determines the solutions based on the solutions calculated by the solution generation unit 15 and a solution presentation model to which the parameters indicated by the parameter information 41 are applied. For example, the solution presentation unit 17 generates multiple sets of candidate solutions and determines the solutions based on the information output by the solution presentation model when these are input into the solution presentation model. The number N may be stored in advance in a storage device 4 or the like, or it may be determined based on user input. The solution presentation unit 17 is an example of a "determination means".

[0033] The UI control unit 18 controls the reception of user input and the display of information to be viewed by the user during the learning phase and presentation phase of the learning unit 16 and the presentation solution determination unit 17, respectively. In this case, the UI control unit 18 receives user input based on the input information S1 supplied from the input device 2 and controls the display of the display device 3 by supplying display information S2 to the display device 3. The specific processing of the UI control unit 18 will be described later with reference to the display example. The UI control unit 18 is an example of a "control means" and a "presentation means".

[0034] The solution generation unit 15, learning unit 16, proposed solution determination unit 17, and UI control unit 18 described in Figure 3 can be realized, for example, by the processor 11 executing a program. Alternatively, the necessary programs may be recorded on any non-volatile storage medium and installed as needed to realize each component. At least a portion of these components may be realized not only by software programs, but also by a combination of hardware, firmware, and software. Furthermore, at least a portion of these components may be realized using user-programmable integrated circuits, such as FPGAs (Field-Programmable Gate Arrays) or microcontrollers. In this case, the program composed of the above components may be realized using this integrated circuit. At least a portion of each component may also be composed of ASSPs (Application Specific Standard Produce), ASICs (Application Specific Integrated Circuits), or quantum processors (quantum computer control chips). Thus, each component may be realized by various hardware. The same applies to other embodiments described later. Furthermore, each of these components may be realized by the collaboration of multiple computers, for example, using cloud computing technology.

[0035] (4) Processing flow Figure 4 is an example of a flowchart executed by the information processing device 1 during the learning phase of the first embodiment.

[0036] First, the solution generation unit 15 generates a predetermined number of solutions to the optimization problem identified by the condition information 40 (step S11). In this case, the predetermined number is set to a value greater than N. Next, the learning unit 16 determines multiple sets of multiple solutions (step S12). In this case, for example, the learning unit 16 generates a predetermined number of sets consisting of N solutions. In this case, the learning unit 16 may generate multiple sets based on any method.

[0037] Next, the UI control unit 18 displays the multiple sets of multiple solutions determined in step S12 on the display device 3 and accepts input to select a set of multiple solutions that has a variation preferred by the user from among the multiple sets of multiple solutions (step S13). Then, when the UI control unit 18 receives input information S1 indicating the user's selection result from the input device 2, it notifies the learning unit 16 of the user's selection result.

[0038] Next, the learning unit 16 updates the parameter information 41 based on the selection results received in step S13 (step S14). In this case, for example, the learning unit 16 considers the selected set of multiple solutions as the correct data for the presented solution and determines the parameters of the presented solution decision model that minimize the error between the output result of the presented solution decision model and the correct data.

[0039] The learning unit 16 then determines whether or not to terminate the learning process (step S15). For example, the learning unit 16 may make the above determination based on whether or not it has executed step S14 a predetermined number of times, or it may make the above determination based on user input. If the learning unit 16 determines that the learning process should be terminated (step S15; Yes), it terminates the flowchart processing. On the other hand, if the learning unit 16 determines that the learning process should not be terminated (step S15; No), it returns to step S11. The learning unit 16 may return to step S12 instead of returning to step S11. In this case, the learning unit 16 executes step S12 based on the solution generated in the already executed step S11.

[0040] Figure 5 is an example of a flowchart executed by the information processing device 1 during the presentation phase of the first embodiment.

[0041] First, the solution generation unit 15 generates a predetermined number of solutions to an optimization problem that is specified based on user input or the like (step S21). In this case, the predetermined number is set to a value greater than N.

[0042] Next, the solution determination unit 17 determines the solution based on the learned parameters indicated by the parameter information 41 (step S22). In this case, the solution determination unit 17 determines the solution based on information output by the solution determination model, which is configured based on the learned parameters, when multiple sets of candidate solutions are input to the solution determination model.

[0043] Then, the UI control unit 18 displays information about the proposed solution determined by the proposed solution determination unit 17 on the display device 3 (step S23). In this case, the UI control unit 18 may highlight the differences between multiple solutions or display a ranking of solutions for each item that is an indicator for evaluating the solution. An example of this display will be described later.

[0044] (5) Display example Figure 6 shows an example of the solution selection screen, which is the screen that the UI control unit 18 displays on the display device 3 in step S13 of Figure 4, which is the learning phase. Based on the processing results of steps S11 and S12, the UI control unit 18 generates display information S2 for displaying the solution selection screen, and transmits the generated display information S2 to the display device 3 via the interface 13, thereby displaying the solution selection screen on the display device 3.

[0045] The solution selection screen has candidate display fields 50 (50A to 50C) that display information about multiple candidate solutions (in this case, Candidates A to C) that the user can select. Here, each candidate represents a set of three solutions, and each candidate display field 50 has a solution display field 51 corresponding to each solution. Each solution has a sequential identification name (Solution 1, Solution 2, ...) assigned in descending order of the value of the objective function used to optimize the optimization problem (objective function value). Here, as an example, a larger objective function value is preferable.

[0046] Each solution display field 51 shows the corresponding solution's identifier, objective function value, and the values ​​of each item. Each solution display field 51 also has a details button to display detailed information about the corresponding solution. Here, the items (item a, item b, ...) are indicators in the optimization problem, and may be indicators used in the objective function or constraint equations, or they may be the variables themselves that are sought as the solution.

[0047] The candidate display fields 50 (50A to 50C) can each be selected based on user input, and the UI control unit 18 prompts the user to select the candidate display field 50 corresponding to the set of solutions with the most favorable variations for the user. When one candidate display field 50 is selected based on a click or touch operation on the solution selection screen, the UI control unit 18 receives input information S1 indicating the selected candidate display field 50 from the input device 2 and supplies the selection result indicated by the input information S1 to the learning unit 16. Subsequently, if the learning phase continues, the UI control unit 18 displays a solution selection screen showing new multiple solution candidates on the display device 3 and accepts the selection of a set of multiple solutions with the most favorable variations for the user.

[0048] Generally, what users prioritize varies from person to person; some prioritize diversity, while others prioritize certain items being within a predetermined range. Taking this into consideration, the information processing device 1 accepts the selection of a set of multiple solutions with variations favorable to the user on the solution selection screen. This allows the information processing device 1 to appropriately obtain data for learning the user's priorities.

[0049] Figure 7 shows an example of the display of the solution presentation screen, which is the screen that the UI control unit 18 displays on the display device 3 in step S23 of Figure 5, which is the presentation phase. Based on the processing results of steps S21 and S22, the UI control unit 18 generates display information S2 for displaying the solution presentation screen, and transmits the generated display information S2 to the display device 3 via the interface 13, thereby displaying the solution presentation screen on the display device 3.

[0050] The UI control unit 18 displays a solution presentation table 53 on the solution presentation screen, which shows the details of each item of the presented solutions (solution X, solution Y, solution Z, ...) determined in step S22. The solution presentation table 53 has a record for each solution, and each record is provided with a detail button 54 for checking the detailed information of the corresponding solution.

[0051] Furthermore, the UI control unit 18 highlights items on the proposed solution table 53 that show significant differences in the proposed solutions. Here, the UI control unit 18 highlights the top two items (item a and item c) with the largest variance (i.e., variation) in the proposed solutions using a border effect. Specifically, the UI control unit 18 surrounds item a, which has the largest variance in the proposed solutions, with a thick border, and surrounds item c, which has the second largest variance, with a thin border. The UI control unit 18 may also determine the number of items to highlight, the type of highlighting (including border thickness, whether or not the background color is changed, etc.) based on setting information generated from user input.

[0052] Here, the solutions displayed in the suggested solutions table 53 are combinations that take into account the trend of the solution sets selected by the user during the learning phase (i.e., the trend of variations that the user values). Therefore, the UI control unit 18 can prevent the display of a large number of solutions in a cumbersome manner and effectively display solutions that are of high importance to the user.

[0053] Furthermore, the UI control unit 18 displays an item-specific ranking display button 55 on the solution presentation screen. When it detects that the item-specific ranking display button 55 has been selected, it displays a screen on the display device 3 that shows the ranking of the presented solutions for each item (also called the "ranking screen"). In this case, preferably, each item in the presented solution table 53 is selectable, and the UI control unit 18 may generate a ranking screen that shows the ranking of the items that were selected on the presented solution table 53 at the time the item-specific ranking display button 55 was selected.

[0054] Figure 8 shows an example of the ranking screen display. When the UI control unit 18 detects, for example, that the item-specific ranking display button 55 has been selected on the solution presentation screen, it transmits display information S2 to the display device 3 via the interface 13, thereby causing the display device 3 to display the ranking screen.

[0055] The UI control unit 18 displays a ranking table 56 on the ranking screen that shows the ranking of the presented solutions for each item. For example, the ranking for item a is in the order of solution X, solution Y, and solution Z. Here, the ranking for each item may be determined based on whether the item is close to a predetermined optimal value (which may be a range, the same applies hereinafter), or it may be determined by further considering the objective function value of the solution in addition to its proximity to the optimal value. In the latter case, the weighting values ​​for proximity to the optimal value and the objective function value can be specified by the user, for example, and the setting values ​​specified by user input are stored in advance in the storage device 4, etc.

[0056] In this way, the UI control unit 18 displays a ranking screen, allowing the user to easily grasp the preferred solution for each item.

[0057] <Second Embodiment> Figure 9 shows the configuration of the optimization system 100A in the second embodiment. As shown in Figure 9, the optimization system 100A mainly comprises an information processing device 1A, a storage device 4, and a terminal device 5. The information processing device 1A and the terminal device 5 communicate data via a network 6.

[0058] The information processing device 1A performs optimization processing. In this case, the information processing device 1A receives input information S1, which the information processing device 1 receives from the input device 2 in the first embodiment, from the terminal device 5 via the network 6. The information processing device 1A also transmits display information S2, which the information processing device 1 transmits to the display device 3 in the first embodiment, to the terminal device 5 via the network 6. Thus, the information processing device 1A according to the second embodiment functions as a server device.

[0059] Terminal device 5 is a terminal having input, display, and communication functions, and functions as the input device 2 and display device 3 in the first embodiment. Terminal device 5 may be, for example, a personal computer, a tablet terminal, or a PDA (Personal Digital Assistant). Terminal device 5 transmits input information S1 generated based on received user input to information processing device 1A via network 6. Also, when terminal device 5 receives display information S2 from information processing device 1A, it displays various screens based on the display information S2.

[0060] The information processing device 1A according to the second embodiment can suitably perform the input processing and display processing performed by the information processing device 1 in the first embodiment for the user of the terminal device 5.

[0061] <Third Embodiment> Figure 10 is a functional block diagram of the learning device 1X in the third embodiment. The learning device 1X mainly comprises a solution generation means 15X and a learning means 16X. The learning device 1X may be composed of multiple devices.

[0062] The solution generation means 15X generates a solution to the optimization problem. The solution generation means 15X can be, for example, the solution generation unit 15 in the first or second embodiment.

[0063] The learning means 16X generates multiple sets of solutions and learns a model that determines the set of solutions to output based on the set selected from the multiple sets by external input. The learning means 16X can be, for example, the learning unit 16 in the first or second embodiment.

[0064] Figure 11 is an example of a flowchart executed by the learning device 1X in the third embodiment. First, the solution generation means 15X generates a solution to the optimization problem (step S31). The learning means 16X generates multiple sets of solutions (step S32). Then, the learning means 16X learns a model that determines the set of solutions to output based on the set selected from the multiple sets by external input (step S33).

[0065] In the third embodiment, the learning device 1X can learn a model to determine a set of solutions that matches the user's preferences when learning a model to determine a set of solutions.

[0066] In addition, some or all of the above embodiments (including modifications, the same applies hereinafter) may also be described as follows, but are not limited to the following.

[0067] [Note 1] Solution generation means for generating solutions to optimization problems, A learning means for generating multiple sets of the aforementioned solutions and learning a model that determines the set of solutions to output based on the set selected from the multiple sets by external input, A learning device having the following features. [Note 2] The learning device according to Appendix 1, further comprising control means for displaying the plurality of sets on a display device and receiving the external input for selecting any of the plurality of sets. [Note 3] The learning device according to Appendix 1 or 2, wherein the learning means repeatedly performs the generation of multiple sets and selection from the multiple sets, and updates the parameters of the model based on each of the sets obtained by the selection. [Note 4] The aforementioned model is a learning device as described in any one of the appendices 1 to 3, which is a learning model that is trained to output information indicating the set of solutions to be presented from among multiple sets of solutions when multiple sets of solutions are input. [Note 5] The set of solutions is presented to the person who made the external input, using the learning device described in any one of the appendices 1 to 4. [Note 6] Solution generation means for generating solutions to optimization problems, A determination means that determines a set of solutions based on a plurality of solutions generated by the solution generation means and a model learned by the learning device described in any one of the appendices 1 to 5, A presentation means for presenting the set of solutions, A display device having the following features. [Note 7] The presentation means is a presentation device as described in Appendix 6, which presents a table showing the values ​​of each solution in the set of solutions for each item. [Note 8] The presentation device described in Appendix 7 determines the items to be highlighted in the table based on the variation in the values ​​of each of the solutions for each of the items. [Note 9] The presentation means is a presentation device according to any one of the appendices 6 to 8, which presents a ranking of the set of solutions for each item. [Note 10] Computers To generate a solution to an optimization problem, Multiple sets of the aforementioned solutions are generated, A model is trained to determine the set of solutions to be output, based on the set selected from the aforementioned multiple sets by external input. Learning methods. [Note 11] To generate a solution to an optimization problem, Multiple sets of the aforementioned solutions are generated, A storage medium containing a program that causes a computer to perform a process of learning a model that determines the set of solutions to be output based on the set selected from the aforementioned multiple sets by external input.

[0068] In each of the embodiments described above, the program can be stored using various types of non-transitory computer-readable medium and supplied to a computer, such as a processor. Non-transitory computer-readable mediums include various types of tangible storage mediums. Examples of non-transitory computer-readable mediums include magnetic storage mediums (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical storage mediums (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, and semiconductor memories (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, and RAMs (Random Access Memory)). The program may also be supplied to the computer by various types of transient computer-readable mediums. Examples of transient computer-readable mediums include electrical signals, optical signals, and electromagnetic waves. Transitory computer-readable mediums can supply the program to the computer via wired communication channels such as electric wires and optical fibers, or via wireless communication channels.

[0069] Although the present invention has been described above with reference to embodiments, the present invention is not limited to the above embodiments. Various modifications to the structure and details of the present invention can be made that are understandable to those skilled in the art within the scope of the present invention. That is, the present invention naturally includes the full disclosure, including the claims, and various modifications and alterations that those skilled in the art could make in accordance with the technical idea. Furthermore, each disclosure of the above-mentioned patent documents and other references is incorporated herein by reference. [Explanation of Symbols]

[0070] 1. 1A Information Processing Device 1X Learning Device 2 Input devices 3 Display device 4 Storage device 5 Terminal devices 100, 100A Optimization System

Claims

1. Solution generation means for generating solutions to optimization problems, A learning means that generates multiple sets of the aforementioned solutions, and for each of the multiple sets, presents the multiple solutions included in the set and the values ​​of the items that are indicators or variables to be obtained as solutions used in the objective function or constraint equation of the optimization problem for each of the multiple solutions, and learns a model that determines the set of the aforementioned solutions to be output based on the set selected from the multiple sets by the user's external input, A learning device having the following features.

2. The learning device according to claim 1, further comprising control means for displaying the plurality of solutions and the value of the item on a display device for each of the plurality of sets, and receiving the external input for selecting any set of the plurality of sets.

3. The learning device according to claim 1 or 2, wherein the learning means repeatedly performs the generation of the plurality of sets and selection from the plurality of sets, and updates the parameters of the model based on each of the sets obtained by the selection.

4. The learning device according to any one of claims 1 to 3, wherein the model is a learning model that is trained to output information indicating a set of solutions to be presented from among multiple sets of solutions when multiple sets of solutions are input.

5. The learning device according to any one of claims 1 to 4, wherein the set of solutions is presented to the person who made the external input.

6. Solution generation means for generating solutions to optimization problems, A determination means for determining a set of solutions based on a plurality of solutions generated by the solution generation means and a model learned by the learning device described in any one of claims 1 to 5, A presentation means for presenting the set of solutions, A display device having the following features.

7. The presentation device according to claim 6, wherein the presentation means presents a table showing the values ​​of items that are indicators used in the objective function or constraints of the optimization problem, or variables to be obtained as solutions, for each solution in the set of solutions.

8. The presentation device according to claim 7, wherein the presentation means determines the items to be highlighted in the table based on the variation in the values ​​of the items in the set of solutions.

9. Computers To generate a solution to an optimization problem, A learning method that generates multiple sets of the aforementioned solutions, presents, for each of the multiple sets, the multiple solutions included in the set and the values ​​of the items that are indicators used in the objective function or constraint equation of the optimization problem or variables to be obtained as solutions for each of the multiple solutions, and learns a model that determines the set of the aforementioned solutions to be output based on the set selected from the multiple sets by the user's external input.

10. To generate a solution to an optimization problem, A program that generates multiple sets of the aforementioned solutions, presents, for each of the multiple sets, the multiple solutions included in that set and the values ​​of the items that are indicators or variables to be obtained as solutions used in the objective function or constraints of the optimization problem for each of the multiple solutions, and causes a computer to perform a process of learning a model that determines the set of solutions to output based on the set selected from the multiple sets by the user's external input.

Images