Information processing apparatus and information processing method

Abstract

To optimize an input parameter in a short time.SOLUTION: An information processing apparatus includes an output data acquisition unit, an output feature amount extraction unit, a first reference feature amount generation unit, a second reference feature amount generation unit, an evaluation value setting unit, an input parameter determination unit, and a repetition determination unit. The output data acquisition unit acquires output data obtained by performing an experiment or simulation based on an input parameter and a physical quantity of the output data. The output feature amount extraction unit inputs the output data to a learned model and extracts a feature amount of the output data. The first reference feature amount generation unit generates a first reference feature amount. The second reference feature amount generation unit generates a second reference feature amount based on the physical quantity of the output data and a degree of similarity between the first reference feature amount and the feature amount of the output data. The evaluation value setting unit sets an evaluation value based on the physical quantity of the output data and a degree of similarity between the second reference feature amount and the feature amount of the output data. The input parameter determination unit determines an input parameter for a next experiment or simulation based on the evaluation value. The repetition determination unit repeats the above processing of each unit until a predetermined condition is satisfied.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] An embodiment of the present invention relates to an information processing device and an information processing method. [Background technology]

[0002] When simulating the behavior of complex events using a simulator, the output data may be multidimensional. When the output data is multidimensional, a huge amount of calculation processing is required to update the input parameters. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7344149 Summary of the Invention [Problem to be solved by the invention]

[0004] An embodiment of the present invention provides an information processing apparatus and an information processing method that can optimize input parameters in a short time. [Means for solving the problem]

[0005] In order to solve the above-mentioned problems, according to one embodiment of the present invention, there is provided a method for generating a physical quantity of an output data by performing an experiment or a simulation based on input parameters, the method comprising: an output feature extraction unit that inputs the output data into a trained model and extracts features of the output data; a first reference feature generating unit that generates a first reference feature; a second reference feature generating unit that generates a second reference feature based on a similarity between the first reference feature and a feature of the output data and a physical quantity of the output data; an evaluation value setting unit that calculates a similarity between the second reference feature amount and the feature amount of the output data, and sets an evaluation value based on the calculated similarity and the physical quantity of the output data; an input parameter determination unit that determines input parameters for a next experiment or simulation based on the evaluation value; An information processing device is provided, comprising: an iterative determination unit that repeats the processing of the output data acquisition unit, the output feature extraction unit, the first reference feature generation unit, the second reference feature generation unit, the evaluation value setting unit, and the next input parameter determination unit until a predetermined condition is satisfied. [Brief explanation of the drawings]

[0006] [Figure 1] FIG. 1 is a block diagram showing the overall configuration of an information processing apparatus according to an embodiment. [Figure 2] FIG. 2 is a diagram illustrating a processing operation of an information processing apparatus according to an embodiment. [Figure 3] 10 is a flowchart showing an example of a processing procedure of a next input parameter determination unit. [Figure 4] FIG. 10 is a diagram illustrating a processing operation of a second reference feature generating unit. [Figure 5] 5 is a diagram showing an example in which each piece of output data input to a second reference feature generating unit in FIG. 4 includes a physical quantity corresponding to a composite output image. [Figure 6] 10A and 10B are diagrams illustrating a processing operation in which a first reference feature generating unit generates a first reference feature without a reference image. [Figure 7] 10A and 10B are diagrams illustrating the processing operation of a feature amount exclusion unit that excludes some feature amounts to generate first and second reference feature amounts. [Figure 8] FIG. 10 is a block diagram showing the overall configuration of an information processing device according to a modified example of an embodiment. [Figure 9] FIG. 10 is a diagram illustrating a processing operation of an information processing device according to a modified example. [Figure 10] 10A and 10B are diagrams illustrating a method for generating a second reference feature based on a contribution rate of a physical quantity. [Figure 11] 4A to 4C are diagrams showing a first example of image synthesis processing performed by an image synthesis unit. [Figure 12] 10A and 10B are diagrams showing a second example of image synthesis processing performed by the image synthesis unit. [Figure 13] 10A and 10B are diagrams showing a third example of image synthesis processing performed by the image synthesis unit. DETAILED DESCRIPTION OF THE INVENTION

[0007] Hereinafter, an embodiment of an information processing device and an information processing method will be described with reference to the drawings. The following description will focus on the main components of the information processing device and the information processing method, but the information processing device and the information processing method may include components and functions that are not shown or described. The following description does not exclude components and functions that are not shown or described.

[0008] (Overall configuration of information processing device) 1 is a block diagram showing the overall configuration of an information processing device 1 according to one embodiment. The information processing device 1 according to the one embodiment includes an output data acquisition unit 2, an output feature extraction unit 3, a first reference feature generation unit 4, a second reference feature generation unit 5, an evaluation value setting unit 6, a next input parameter determination unit 7, and a repetition determination unit 8.

[0009] The information processing device 1 according to one embodiment is characterized in that it determines input parameters for, for example, a simulator 9 or an experimental device. The content of the simulation or experiment performed by the simulator 9 or the experimental device is not important. For example, the structure, surface shape, or electrical characteristics of a semiconductor device manufactured by a semiconductor manufacturing device may be simulated or experimented.

[0010] The output data acquisition unit 2 acquires output data obtained by conducting an experiment or simulation based on the input parameters, and physical quantities of the output data. The output data acquisition unit 2 acquires one or more input parameters. The input parameters are various parameters such as pressure, voltage, temperature, and process conditions. The physical quantities are values or quantities of pressure, voltage, temperature, process conditions, etc. In a more specific example, the physical quantities include the flow rate or concentration of a processing medium used in a manufacturing apparatus that manufactures the processing object. The specific types and numbers of the input parameters and physical quantities are not important.

[0011] The output feature extraction unit 3 inputs output data including a synthesized output image synthesized by the image synthesis unit 11 into the trained model 10 or output data acquired by the output data acquisition unit 2 and stored in the input / output data storage unit 12, and extracts features of the output data. The output data includes, for example, an output image obtained by the simulator 9 or an experiment. The model 10 outputs features of the input data. The model 10 is, for example, one of various network models capable of performing machine learning, such as a convolutional neural network (CNN) or a generative adversarial network (GAN). The detailed structure of the model 10 is not important. The model 10 may be provided outside the information processing device 1. In this case, the features output from the model 10 are input to the information processing device 1 as external input data. The following mainly describes an example in which the output data includes an output image and the output feature extraction unit 3 extracts features of the output image.

[0012] The first reference feature generation unit 4 generates the first reference feature. As will be described later, the first reference feature generation unit 4 generates the first reference feature based on a predetermined reference image and physical quantities of output data (e.g., output images) while taking into account user knowledge. Alternatively, the first reference feature generation unit 4 generates the first reference feature based on feature quantities of one or more output data corresponding to physical quantities that satisfy a predetermined criterion among physical quantities of multiple output data (e.g., output images) without using a reference image. In a more specific example, the first reference feature generation unit 4 may set the center of gravity of feature quantities of two or more output images corresponding to physical quantities that satisfy a predetermined condition as the first reference feature. Alternatively, the first reference feature generation unit 4 may generate the first reference feature by averaging feature quantities of two or more output data corresponding to physical quantities that satisfy a predetermined condition. Alternatively, the first reference feature generation unit 4 may set the feature quantities of output data corresponding to physical quantities that satisfy a predetermined criterion as the first reference feature.

[0013] The second reference feature generating unit 5 calculates the similarity between the first reference feature and the feature of output data (for example, an output image), and generates the second reference feature based on this similarity and the physical quantity of the output data.

[0014] The evaluation value setting unit 6 calculates the similarity between the second reference feature and the feature of the output data, and sets an evaluation value based on the calculated similarity and the physical quantity of the output data. The evaluation value indicates a combination of the similarity and the corresponding physical quantity.

[0015] The next input parameter determination unit 7 determines input parameters for the next experiment or simulation based on the evaluation value.

[0016] The iteration determination unit 8 repeats the processing of the output data acquisition unit 2, the output feature extraction unit 3, the first reference feature generation unit 4, the second reference feature generation unit 5, the evaluation value setting unit 6, and the next input parameter determination unit 7 until a predetermined condition is satisfied. Satisfying the predetermined condition may mean, for example, that the number of times an experiment or simulation has been executed reaches a predetermined limit, that the elapsed time since the start of the experiment or simulation exceeds a predetermined limit, or that the above-mentioned evaluation value reaches a predetermined threshold.

[0017] The information processing device 1 according to an embodiment may include an image synthesis unit 11. The image synthesis unit 11 classifies a plurality of output images into a plurality of channels, and then synthesizes the output images of each channel to generate a synthesized output image. The channels are assigned to different processing objects, for example. As will be described later, the image synthesis unit 11 may perform grayscale conversion on the plurality of output images to reduce the amount of data, and then synthesize the plurality of output images. Alternatively, the image synthesis unit 11 may synthesize a plurality of output images generated by changing the weight for each of the plurality of channels to generate a synthesized output image.

[0018] The information processing device 1 according to an embodiment may include an input / output data storage unit 12. The input / output data storage unit 12 stores a pair of output data acquired by the output data acquisition unit 2 and input parameters corresponding to the output data. Since the output data includes an output image and a physical quantity, the input / output data storage unit 12 stores a pair of the output image, the physical quantity, and the input parameters. The output data stored in the input / output data storage unit 12 is input to the image synthesis unit 11. The output data stored in the input / output data storage unit 12 may also be input to the output feature extraction unit 3 and the evaluation value setting unit 6. A composite output image synthesized by the image synthesis unit 11 may also be input to the output feature extraction unit 3.

[0019] 2 is a diagram illustrating the processing operation of the information processing device 1 according to an embodiment of the present invention, showing a process for optimizing input parameters of a composite output image synthesized by the image synthesis unit 11.

[0020] The first reference feature generator 4 inputs, for example, a reference image into a trained model 10, and generates a first reference feature based on the feature f0 of the reference image output from the model 10 and the physical quantity of the composite output image. The reference image is read out, for example, from a past knowledge database (past knowledge DB) 13.

[0021] The output feature extraction unit 3 inputs the composite output image to the trained model 10 and extracts the feature of the composite output image.

[0022] The second reference feature generating unit 5 generates the second reference feature based on the similarity between the first reference feature and the feature of the composite output image, and the physical quantity of the composite output image.

[0023] The evaluation value setting unit 6 compares the second reference feature amount with the feature amount of the composite output image to calculate the similarity between the second reference feature amount and the composite output image. Next, the evaluation value setting unit 6 sets an evaluation value that is a combination of the similarity and the physical quantity of the composite output image.

[0024] The next input parameter determination unit 7 determines input parameters for the next experiment or simulation based on the evaluation value. The determined input parameters are input to the experimental device or simulator 9.

[0025] The output data acquisition unit 2 acquires output data indicating the experimental results or simulation results of the experimental device or simulator 9 and the input parameters determined by the next input parameter determination unit 7 in association with each other.

[0026] (Procedure for determining the next input parameter) Fig. 3 is a flowchart showing an example of a processing procedure of the next input parameter determination unit 7. The flowchart in Fig. 3 shows an example of Bayesian optimization. First, the relationship between the input parameters and the similarity and physical quantities is estimated by a Gaussian process (step S1).

[0027] Next, based on the estimation result, the similarity in the evaluation value and the acquisition function of the input parameter candidate data for the physical quantity of the composite output image are calculated (step S2). Next, the sum of the acquisition functions is calculated for each input parameter candidate data (step S3). Next, the input parameter candidate data that maximizes the sum of the acquisition functions is determined as the next input parameter (step S4).

[0028] (Processing operation of second reference feature generating unit 5) 4 is a diagram illustrating the processing operation of the second reference feature generating unit 5. An experimental device or simulator 9 outputs a plurality of output data D1 to Dn. Each of the output data D1 to Dn may include physical quantities corresponding to one output image, or may include physical quantities corresponding to a composite output image obtained by combining a plurality of output images.

[0029] FIG. 5 is a diagram showing an example in which each of the output data D1 to Dn input to the second reference feature generating unit 5 in FIG. 4 includes a composite output image and corresponding physical quantities. FIG. 5 shows the data structure of the output data D1, but the output data D2 to Dn have the same data structure. In the example of FIG. 5, the output data D1 includes three output images. The three output images, for example, each have a different type of information. The type of information indicates, for example, the distribution of a different physical quantity. More specifically, the type of information includes, for example, a flow velocity map and a concentration map of the processing object. In this case, the output image includes an image of the flow velocity map and an image of the concentration map.

[0030] As shown in FIG. 5, when each of the output data D1 to Dn includes a physical amount corresponding to the composite output image, the second reference feature generating unit 5 calculates the similarity between the feature amount of the composite output image and the first reference feature amount.

[0031] In addition, the second reference feature generating unit 5 generates the second reference feature based on the similarity between the first reference feature and the feature of the composite output image and the physical quantity of the composite output image that satisfies a predetermined condition.

[0032] Taking the magnitude relationship of physical quantities as an example, the predetermined condition is that the physical quantity is equal to or less than 3. In this case, the second reference feature generation unit 5 generates the second reference feature taking into consideration the similarity between the feature of each output image having a physical quantity equal to or less than 3 and the first reference feature.

[0033] 4 and 5 show an example in which a composite output image is input to the second reference feature generation unit 5, but an output image that has not been composited may also be input to the second reference feature generation unit 5.

[0034] (Generating the first reference feature without user knowledge or reference images) The first reference feature generation unit 4 can generate first reference features from a reference image, and can also generate first reference features without user knowledge and without using a reference image. Fig. 6 is a diagram illustrating the processing operation of the first reference feature generation unit 4 to generate first reference features without a reference image.

[0035] The first reference feature generator 4 sets the center of gravity of two or more feature quantities obtained by inputting, into the trained model 10, two or more output images corresponding to physical quantities that satisfy a predetermined condition, among a plurality of physical quantities included in a plurality of output data output from, for example, an experimental device or a simulator 9, as the first reference feature quantity. For example, the predetermined condition is a condition that the physical quantity is 3 or less, taking the magnitude relationship of the physical quantities as an example. Instead of calculating the center of gravity of two or more feature quantities, the average value of the two or more feature quantities may be set as the first reference feature quantity.

[0036] This makes it possible to generate the first reference feature amount without requiring the user's knowledge or preparing a reference image in advance, thereby saving the user time and effort.

[0037] (The first and second reference features are generated by excluding some features.) When generating at least one of the first reference feature amount and the second reference feature amount, at least one of the first reference feature amount and the second reference feature amount can be generated after excluding some feature amounts.

[0038] FIG. 7 is a diagram illustrating the processing operation of the feature amount excluding unit 14 that generates the first reference feature amount and the second reference feature amount by excluding some feature amounts.

[0039] The feature excluding unit 14 specifies in advance feature quantities that are not preferable as the first reference feature quantities or the second reference feature quantities, and sets the specified feature quantities as excluded feature quantities. The feature excluding unit 14 calculates the similarity between the excluded feature quantities and multiple feature quantities obtained by inputting multiple output images included in multiple output data output from, for example, an experimental device or a simulator 9 into the trained model 10, and excludes feature quantities with high similarity that exceeds a predetermined threshold.

[0040] At least one of the first reference feature generating unit 4 and the second reference feature generating unit 5 generates at least one of the first reference feature and the second reference feature from feature amounts other than the feature amounts excluded by the feature excluding unit 14.

[0041] In this way, by providing the feature amount removal unit 14, it is possible to remove inappropriate feature amounts in advance, and the processing by the first reference feature amount generation unit 4 and the second reference feature amount generation unit 5 can be performed efficiently.

[0042] (Change in contribution rate of physical quantities) When a plurality of output data items including a plurality of output images and a plurality of corresponding physical quantities are input to the output feature extraction unit 3, a contribution rate can be set for each of the plurality of physical quantities, and the usage rate of each physical quantity can be updated for each experiment or simulation according to the contribution rate of each physical quantity.

[0043] FIG. 8 is a block diagram showing the overall configuration of an information processing device 1 according to a modification of the embodiment, and FIG. 9 is a diagram illustrating the processing operation of the information processing device 1 according to the modification.

[0044] As shown in FIGS. 8 and 9, the information processing device 1 according to the modification includes a similarity storage unit 15 and a physical quantity contribution rate calculation unit 16 in addition to the configuration shown in FIGS.

[0045] An evaluation value setting unit 6 calculates multiple similarities between each of the feature quantities of the multiple output images extracted by the output feature quantity extraction unit 3 and the second reference feature quantity. A similarity storage unit 15 stores the calculated similarities in association with the variations in the physical quantities corresponding to each output image. A physical quantity contribution rate calculation unit 16 calculates the contribution rates of the multiple physical quantities based on the calculated similarities and the variations in the physical quantities corresponding to the multiple output images. For example, the greater the variation in a physical quantity, the lower the calculated contribution rate.

[0046] The first reference feature generating unit 4 generates a first reference feature by using each physical quantity at a usage rate according to the contribution rate of each physical quantity calculated by the physical quantity contribution rate calculating unit 16.

[0047] The second reference feature quantity generation unit 5 uses each physical quantity at a usage rate according to the contribution rate of each physical quantity calculated by the physical quantity contribution rate calculation unit 16, and generates the second reference feature quantity based on the similarity between the first reference feature quantity and the feature quantity of each output image. Here, as the contribution rate of each physical quantity, the contribution rate used by the first reference feature quantity generation unit 4 may be used as is, or a new contribution rate input from the physical quantity contribution rate calculation unit 16 may be used.

[0048] 10 is a diagram illustrating a method for generating second reference features based on the contribution rates of physical quantities. The similarity storage unit 15 stores pairs of the variations in each physical quantity included in the output data of the experimental device or simulator 9 obtained in the process of past optimization processing of the input parameters and the similarities calculated by the evaluation value setting unit 6.

[0049] The physical quantity contribution rate calculation unit 16 defines the contribution rate of each physical quantity based on the similarity stored in the similarity storage unit 15 and the variation of each physical quantity.

[0050] Thereafter, when new output data is output from the simulator 9 or the experimental device, the physical quantity contribution rate calculation unit 16 determines the contribution rate of each physical quantity by using the variation of each physical quantity corresponding to each output image included in each output data.

[0051] The second reference feature generating unit 5 changes the usage ratio of each physical quantity based on the contribution rate of each physical quantity determined by the physical quantity contribution rate calculating unit 16, and generates a second reference feature based on the above-mentioned similarity.

[0052] (First example of image synthesis) The image synthesis unit 11 can classify a plurality of output images included in a plurality of output data into different channels, perform image processing on the images, and then synthesize the images to generate a synthesized output image.

[0053] FIG. 11 is a diagram showing a first example of image synthesis processing performed by the image synthesis unit 11. An output data set including one or more output data is input to the image synthesis unit 11, for example, from the input / output data storage unit 12 of FIG. 1. FIG. 11 shows one output data D included in the output data set. The output data D includes a plurality of output images and a plurality of physical quantities. The physical quantities are provided in correspondence with each of the plurality of output images. The plurality of output images included in the output data D are, for example, output images relating to a plurality of processing objects manufactured in one lot by a manufacturing device.

[0054] In the first example, the plurality of output images are classified into different channels ch1 to chm. The image composition unit 11 generates a composite output image by combining the output images classified into each of the channels ch1 to chm while maintaining information on the connection between the output images. The generated composite output image is stored in the composite image storage unit 17.

[0055] The output feature quantity extraction unit 3 shown in FIG. 1 extracts the feature quantities of the composite output image stored in the composite image storage unit 17 .

[0056] 12 is a diagram showing the image synthesis process according to the second example performed by the image synthesis unit 11. The image synthesis unit 11 in FIG.

[0057] The evaluation index comparison unit 18 compares the physical quantity corresponding to each output image with a predetermined evaluation index for the physical quantity. For example, the evaluation index comparison unit 18 calculates the square error between the physical quantity and the corresponding evaluation index. Note that the physical quantity and the evaluation index may be compared using a method other than the square error.

[0058] The weight calculation unit 19 calculates a weight for each channel based on the comparison result between the physical quantity and the corresponding evaluation index. In a more specific example, the weight calculation unit 19 calculates a weight for each channel so that the square error between the physical quantity and the corresponding evaluation index is minimized.

[0059] The image synthesis unit 11 adjusts the resolution, size, brightness, color, etc. of the output image of each channel according to the magnitude of the weight, generates a new output image for each channel, and synthesizes the generated new output images to generate a synthesized output image.

[0060] FIG. 13 is a diagram showing a third example of image synthesis processing performed by the image synthesis unit 11. In the third example, the output data set includes multiple output data. Each output data includes multiple output images and multiple physical quantities. All of the multiple output data included in the output data set are, for example, output data related to multiple processing objects manufactured in the same lot. As shown in FIG. 13, the multiple output images are classified into a first direction X and a second direction Y. The second direction Y represents the type of channel. Here, each channel corresponds to one processing object. Multiple output images related to different processing objects are classified into different channels. Multiple output images A to N related to different information about the same processing object are arranged in the first direction X.

[0061] In the third example, each processing object is classified into a separate channel, and a plurality of output images A to N arranged in the first direction X are generated for each processing object. The plurality of output images for each of the plurality of processing objects are all combined to generate a composite output image. The composite output image may be a single image, or may be a bundle of multiple output images.

[0062] The multiple output images arranged in the first direction X are images of the same processing object but each containing different types of information. For example, output image A represents an image of a flow velocity map, and output image B represents an image of a concentration map.

[0063] The evaluation index comparison unit 18 compares the evaluation indexes A to N set for each of the output images A to N with each physical quantity. For example, the evaluation index comparison unit 18 calculates the squared error between the evaluation index A and each of the physical quantities A1 to Am. Similarly, the evaluation index comparison unit 18 calculates the squared error between the evaluation index A and the corresponding physical quantity for each of the evaluation indexes B to N. The weight calculation unit 19 calculates weights W1 to Wm for each channel based on the multiple squared errors calculated by the evaluation index comparison unit 18. The image synthesis unit 11 generates a new output image for each channel based on the weights W1 to Wm calculated by the weight calculation unit 19, and synthesizes the generated new output images to generate a synthesized output image.

[0064] According to the third example, the amount of information in the composite output image can be increased more than in the first and second examples, and a wider range of feature amounts can be extracted.

[0065] In this manner, in this embodiment, a first reference feature is generated based on a physical quantity corresponding to an output image, a second reference feature is generated based on the first reference feature and the feature of the output image, a similarity between the second reference feature and the feature of the output image is calculated, and input parameters for the next experiment or simulation are determined based on an evaluation value that combines the similarity and the physical quantity of the output image, thereby optimizing the input parameters.

[0066] According to this embodiment, the first reference feature can be generated based on the user's knowledge, and the input parameters can be determined taking the user's knowledge into consideration. On the other hand, in this embodiment, the reference image and the user's knowledge are not necessarily required to generate the first reference feature, which saves the user effort and allows the first reference feature to be generated through an automated process. Furthermore, a composite output image can be used when extracting the feature of the output image. The composite output image can contain various information, and the input parameters can be optimized using the composite output image containing information on complex physical quantities. This allows the optimization process to be performed in a short time, even for multidimensional input parameters.

[0067] At least a part of the information processing device 1 described in the above embodiment may be configured with hardware or software. If configured with software, a program that realizes at least a part of the functions of the information processing device 1 may be stored on a recording medium such as a flexible disk or a CD-ROM, and may be read and executed by a computer. The recording medium is not limited to removable recording media such as magnetic disks and optical disks, but may also be fixed recording media such as hard disk drives and memories.

[0068] In addition, a program that realizes at least some of the functions of the information processing device 1 may be distributed via a communication line (including wireless communication) such as the Internet. Furthermore, the program may be encrypted, modulated, or compressed and distributed via a wired line or wireless line such as the Internet, or stored on a recording medium.

[0069] [Note] [Item 1] an output data acquisition unit that acquires output data obtained by performing an experiment or a simulation based on input parameters and a physical quantity of the output data; an output feature extraction unit that inputs the output data into a trained model and extracts features of the output data; a first reference feature generating unit that generates a first reference feature; a second reference feature generating unit that generates a second reference feature based on a similarity between the first reference feature and a feature of the output data and a physical quantity of the output data; an evaluation value setting unit that calculates a similarity between the second reference feature amount and the feature amount of the output data, and sets an evaluation value based on the calculated similarity and the physical quantity of the output data; a next input parameter determination unit that determines input parameters for a next experiment or simulation based on the evaluation value; an iteration determination unit that repeats the processes of the output data acquisition unit, the output feature extraction unit, the first reference feature generation unit, the second reference feature generation unit, the evaluation value setting unit, and the next input parameter determination unit until a predetermined condition is satisfied, Information processing device. [Item 2] the second reference feature generating unit generates the second reference feature based on a similarity between the first reference feature and a feature of the output data and a physical quantity of the output data that satisfies a predetermined condition. Item 1. An information processing device according to item 1. [Item 3] The physical quantity includes a flow rate or a concentration of a processing medium used in a manufacturing apparatus for manufacturing the processing object. Item 3. The information processing device according to item 1 or 2. [Item 4] the output data includes an output image and the physical quantity corresponding to the output image; Item 3. The information processing device according to item 1 or 2. [Item 5] an image synthesis unit that generates a plurality of output images for each of a plurality of channels corresponding to a plurality of processing objects, and then generates a synthesized output image by synthesizing the plurality of output images; Item 5. An information processing device according to item 4. [Item 6] the image synthesis unit performs grayscale conversion on the plurality of output images to reduce the amount of data, and then performs image processing corresponding to each of the plurality of channels. Item 5. An information processing device according to item 5. [Item 7] the image synthesis unit generates the plurality of output images by changing a weight for each of the plurality of channels. Item 7. The information processing device according to item 5 or 6. [Item 8] a weight calculation unit that calculates a plurality of weights corresponding to the plurality of channels based on the plurality of physical quantities corresponding to the plurality of output images and a predetermined evaluation index; the image synthesis unit generates the plurality of output images based on the plurality of weights. Item 8. An information processing device according to item 7. [Item 9] the weight calculation unit calculates the weights so as to minimize a square error between each of the physical quantities and the evaluation index. Item 9. An information processing device according to item 8. [Item 10] the plurality of output images are images of different processing objects, 10. The information processing device according to any one of items 5 to 9. [Item 11] each of the plurality of output images includes a plurality of images relating to different types of information about the corresponding processing object; Item 11. An information processing device according to item 10. [Item 12] the plurality of images are images representing distributions of different physical quantities, Item 12. The information processing device according to item 11. [Item 13] the image synthesis unit classifies the output images corresponding to the processing objects included in one lot into different channels, respectively. 13. The information processing device according to any one of items 10 to 12. [Item 14] the output feature extraction unit inputs the composite output image to the model and extracts features of the composite output image; the evaluation value setting unit calculates the similarity between the second reference feature amount and the feature amount of the composite output image, and sets the evaluation value of the composite output image based on the calculated similarity and a physical quantity of the composite output image. Item 14. The information processing device according to any one of items 5 to 13. [Item 15] the first reference feature generating unit generates the first reference feature from a reference image based on a user's knowledge; 15. The information processing device according to any one of items 4 to 14. [Item 16] the first reference feature generating unit generates the first reference feature based on one or more feature of the output image corresponding to a physical quantity that satisfies a predetermined condition, among the plurality of physical quantities corresponding to the plurality of output images; 15. The information processing device according to any one of items 4 to 14. [Item 17] a contribution rate calculation unit that calculates contribution rates of the plurality of physical quantities based on the plurality of similarities calculated by the evaluation value setting unit and variations in the plurality of physical quantities corresponding to the plurality of output images, the second reference feature generating unit generates the second reference feature based on the plurality of physical quantities having usage rates set based on contribution rates of the plurality of physical quantities and the plurality of similarities. 17. The information processing device according to any one of items 4 to 16. [Item 18] an excluded feature value setting unit that sets a feature value to be excluded; at least one of the first reference feature generation unit and the second reference feature generation unit generates at least one of the first reference feature and the second reference feature, excluding the feature set by the excluded feature setting unit; 18. The information processing device according to any one of items 1 to 17. [Item 19] an input / output data storage unit that stores the output data acquired by the output data acquisition unit, a physical quantity of the output data, and the input parameter corresponding to the output data as a set; 19. The information processing device according to any one of items 1 to 18. [Item 20] On the computer, acquiring output data obtained by performing an experiment or a simulation based on input parameters and physical quantities of the output data; inputting the output data into a trained model and extracting features of the output data; generating a first reference feature; generating a second reference feature based on a similarity between the first reference feature and the feature of the output data and a physical quantity of the output data; calculating a similarity between the second reference feature amount and the feature amount of the output data, and setting an evaluation value based on the calculated similarity and the physical amount of the output data; determining input parameters for a next experiment or simulation based on the evaluation value; repeating the steps of acquiring the output data and the physical quantity of the output data, extracting a feature quantity of the output data, generating the first reference feature quantity, generating the second reference feature quantity, setting the evaluation value, and determining the input parameters until a predetermined condition is satisfied; Information processing methods.

[0070] The aspects of the present disclosure are not limited to the individual embodiments described above, but include various modifications that may be conceived by those skilled in the art, and the effects of the present disclosure are not limited to the above-described contents. In other words, various additions, modifications, and partial deletions are possible within the scope of the conceptual idea and spirit of the present disclosure, which is derived from the contents defined in the claims and their equivalents. [Explanation of symbols]

[0071] 1 Information processing device, 2 Output data acquisition unit, 3 Output feature extraction unit, 4 First reference feature generation unit, 5 Second reference feature generation unit, 6 Evaluation value setting unit, 7 Primary input parameter determination unit, 8 Iterative judgment unit, 9 Simulator, 10 Model, 11 Image synthesis unit, 12 Input / output data storage unit, 13 Past knowledge database (past knowledge DB), 14 Feature exclusion unit, 15 Similarity storage unit, 16 Physical quantity contribution rate calculation unit, 17 Synthesized image storage unit, 18 Evaluation index comparison unit, 19 Weight calculation unit

Claims

1. an output data acquisition unit that acquires output data obtained by performing an experiment or a simulation based on input parameters and a physical quantity of the output data; an output feature extraction unit that inputs the output data into a trained model and extracts features of the output data; a first reference feature generating unit that generates a first reference feature; a second reference feature generating unit that generates a second reference feature based on a similarity between the first reference feature and a feature of the output data and a physical quantity of the output data; an evaluation value setting unit that calculates a similarity between the second reference feature amount and the feature amount of the output data, and sets an evaluation value based on the calculated similarity and the physical quantity of the output data; a next input parameter determination unit that determines input parameters for a next experiment or simulation based on the evaluation value; an iteration determination unit that repeats the processes of the output data acquisition unit, the output feature extraction unit, the first reference feature generation unit, the second reference feature generation unit, the evaluation value setting unit, and the next input parameter determination unit until a predetermined condition is satisfied, Information processing device.

2. the second reference feature generating unit generates the second reference feature based on a similarity between the first reference feature and a feature of the output data and a physical quantity of the output data that satisfies a predetermined condition. The information processing device according to claim 1 .

3. The physical quantity includes a flow rate or a concentration of a processing medium used in a manufacturing apparatus for manufacturing the processing object. The information processing device according to claim 1 .

4. the output data includes an output image and the physical quantity corresponding to the output image; The information processing device according to claim 1 .

5. an image synthesis unit that generates a plurality of output images for each of a plurality of channels corresponding to a plurality of processing objects, and then generates a synthesized output image by synthesizing the plurality of output images; The information processing device according to claim 4 .

6. the image synthesis unit performs grayscale conversion on the plurality of output images to reduce the amount of data, and then performs image processing corresponding to each of the plurality of channels. The information processing device according to claim 5 .

7. the image synthesis unit generates the plurality of output images by changing a weight for each of the plurality of channels. The information processing device according to claim 5 .

8. a weight calculation unit that calculates a plurality of weights corresponding to the plurality of channels based on the plurality of physical quantities corresponding to the plurality of output images and a predetermined evaluation index; the image synthesis unit generates the plurality of output images based on the plurality of weights. The information processing device according to claim 7 .

9. the weight calculation unit calculates the weights so as to minimize a square error between each of the physical quantities and the evaluation index. The information processing device according to claim 8 .

10. the plurality of output images are images of different processing objects, The information processing device according to claim 5 .

11. each of the plurality of output images includes a plurality of images relating to different types of information about the corresponding processing object; The information processing device according to claim 10.

12. the plurality of images are images representing distributions of different physical quantities, The information processing device according to claim 11.

13. the image synthesis unit classifies the output images corresponding to the processing objects included in one lot into different channels, The information processing device according to claim 10.

14. the output feature extraction unit inputs the composite output image to the model and extracts features of the composite output image; the evaluation value setting unit calculates the similarity between the second reference feature amount and the feature amount of the composite output image, and sets the evaluation value of the composite output image based on the calculated similarity and a physical quantity of the composite output image. The information processing device according to claim 5 .

15. the first reference feature generating unit generates the first reference feature from a reference image based on a user's knowledge; The information processing device according to claim 4 .

16. the first reference feature generating unit generates the first reference feature based on feature of one or more of the output images corresponding to physical quantities that satisfy a predetermined condition, among the plurality of physical quantities corresponding to the plurality of output images; The information processing device according to claim 4 .

17. a contribution rate calculation unit that calculates contribution rates of the plurality of physical quantities based on the plurality of similarities calculated by the evaluation value setting unit and variations in the plurality of physical quantities corresponding to the plurality of output images, the second reference feature generating unit generates the second reference feature based on the plurality of physical quantities having usage rates set based on contribution rates of the plurality of physical quantities and the plurality of similarities; The information processing device according to claim 4 .

18. an excluded feature value setting unit that sets a feature value to be excluded; at least one of the first reference feature generation unit and the second reference feature generation unit generates at least one of the first reference feature and the second reference feature, excluding the feature set by the excluded feature setting unit; The information processing device according to claim 1 .

19. an input / output data storage unit that stores the output data acquired by the output data acquisition unit, a physical quantity of the output data, and the input parameter corresponding to the output data as a set; The information processing device according to claim 1 .

20. On the computer, acquiring output data obtained by performing an experiment or a simulation based on input parameters and physical quantities of the output data; inputting the output data into a trained model and extracting features of the output data; generating a first reference feature; generating a second reference feature based on a similarity between the first reference feature and the feature of the output data and a physical quantity of the output data; calculating a similarity between the second reference feature amount and the feature amount of the output data, and setting an evaluation value based on the calculated similarity and the physical quantity of the output data; determining input parameters for a next experiment or simulation based on the evaluation value; repeating the steps of acquiring the output data and the physical quantity of the output data, extracting a feature quantity of the output data, generating the first reference feature quantity, generating the second reference feature quantity, setting the evaluation value, and determining the input parameters until a predetermined condition is satisfied; Information processing methods.

Images