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

The information processing device addresses the inefficiencies of conventional interpolation methods by using a trained model to calculate feature quantities and interpolate missing data, achieving accurate and efficient results with reduced computational load.

WO2024242197A1PCT designated stage expired Publication Date: 2024-11-28PREFERRED NETWORKS INC
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
PCT/JP2024/019244
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-25
Filing Date
2024-05-24
Publication Date
2024-11-28

AI Technical Summary

Technical Problem

Conventional interpolation methods for missing data, such as using average or minimum values, are inadequate when there is a large discrepancy in data averages before and after interpolation, and methods like multiple regression or tensor decomposition are inefficient due to high calculation loads.

Method used

An information processing device that uses a processor to detect missing data, calculates feature quantities, and employs a trained model, such as a deep neural network, to interpolate missing values, reducing computational load and improving accuracy by generating an interpolated data group suitable for further processing.

Benefits of technology

This approach allows for more accurate and efficient interpolation of missing data, reducing data characteristic deviations and optimizing processing, especially in real-time applications with randomly missing values, while minimizing computational load compared to existing methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure JP2024019244_28112024_PF_FP_ABST
    Figure JP2024019244_28112024_PF_FP_ABST
Patent Text Reader

Abstract

An information processing device (1) according to an embodiment comprises at least one memory (33, 35) and at least one processor (31). The at least one processor (31) inputs, into a first model, a plurality of feature amounts calculated from a first data group from which at least one value is missing and which is output from a detection device, and outputs at least one interpolation value used for interpolation of the first data group from which at least one value is missing.
Need to check novelty before this filing date? Find Prior Art

Description

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

[0001] The embodiments of the present disclosure relate to an information processing device, an information processing program, and an information processing method.

[0002] This application is a patent application filed in fiscal year 2023 by the Ministry of Internal Affairs and Communications for the "Research and Development of Elemental Technologies for Providing User-Optimized Data Using Remote Sensing Technology" and is subject to Article 17 of the Industrial Technology Enhancement Act.

[0003] For example, if there is missing data in the detection data obtained using a specific detection device, the detection data cannot be input directly into an information processing device. Therefore, when there is missing data, the missing data (values) are often interpolated using average values ​​or minimum values. In some cases, interpolation is performed using a multiple regression model or tensor decomposition.

[0004] JP 2018-068752 A Japanese Patent No. 7004844 A Japanese Patent No. 6994812 A

[0005] However, interpolation methods that substitute average or minimum values ​​for missing data are not suitable for missing data where the average values ​​of the data before and after interpolation differ significantly. Interpolation methods that use multiple regression models are not suitable when missing data occur uniformly across space. Furthermore, interpolation methods that use tensor decomposition have a high computational load.

[0006] Therefore, there is room for improvement in conventional methods for interpolating missing data.

[0007] The problem to be solved by the present disclosure is to provide a method for interpolating missing data that is more appropriate than conventional methods.

[0008] An information processing device according to an embodiment includes at least one memory and at least one processor, and the at least one processor inputs a plurality of feature quantities calculated from a first data group output from a detection device and having at least one missing value into a first model, and outputs at least one interpolated value used to interpolate the first data group having at least one missing value.

[0009] FIG. 1 is a conceptual diagram illustrating information processing executed by an information processing device according to the present embodiment. FIG. 2 is a block diagram illustrating an example of the hardware configuration of the information processing device according to the embodiment. FIG. 3 is a diagram illustrating an example of functional blocks in a processor according to the embodiment. FIG. 4 is a diagram illustrating the flow of data between functional blocks when inference processing is executed. FIG. 5 is a flowchart illustrating an example of the flow of learning processing of a missing value interpolation model. FIG. 6 is a flowchart illustrating an example of the flow of inference processing executed by the information processing device.

[0010] Hereinafter, an information processing device, an information processing program, and an information processing method according to embodiments will be described in detail with reference to the drawings. In the following description, components having substantially the same functions and configurations are designated by the same reference numerals, and redundant explanations will be provided only when necessary. In addition, the embodiments can be combined with other embodiments, modifications, and conventional techniques.

[0011] The information processing device 1 according to the embodiment may be a device that performs compression processing, data cleansing processing, etc. (hereinafter also referred to as "primary task") on a group of huge amounts of data (hereinafter also referred to as "detection data") acquired by a detection device or an imaging device, such as satellite photos or weather radar, when the group (hereinafter also referred to as "detection data group") is transferred to an external device connected via a network. Note that, for the sake of concrete explanation, the following will be exemplified by a case where the primary task processing is data compression processing.

[0012] The detection data group that is the target of the primary task may include data defects that occur uniformly and randomly in space and time, or data defects that occur non-uniformly in space and time. The information processing device 1 may be a device that performs missing value interpolation processing on such a detection data group and then executes primary task processing. Furthermore, the information processing device 1 may be a device that executes learning processing on a trained model that executes missing value interpolation processing (missing value interpolation model) and a trained model that executes primary task processing (primary task model).

[0013] In the following, for the sake of concreteness, a case where the primary task processing is data compression processing will be taken as an example. Furthermore, the detected data group including missing data before the missing value interpolation processing will also be referred to as the "uninterpolated analysis target data group," and the detected data group after the missing value interpolation processing will also be referred to as the "interpolated analysis target data group."

[0014] 1 is a conceptual diagram illustrating information processing executed by an information processing device 1 according to this embodiment. The information processing device 1 receives a group of uninterpolated analysis target data from a detection device, performs missing value interpolation processing to generate an interpolated analysis target data group. The information processing device 1 executes a primary task process (compression processing) using the interpolated analysis target data group, and transfers the resulting compressed data to an external device, for example, via a network.

[0015] 2 is a block diagram showing an example of a hardware configuration of the information processing device 1 according to the embodiment. As shown in FIG. 2, the information processing device 1 may be connected to an external device 9A via a communication network 5. The information processing device 1 may also include an external device 9B connected via a device interface 39.

[0016] The information processing device 1 may include a computer 30 and an external device 9B connected to the computer 30 via a device interface 39. The computer 30 may include, for example, a processor 31, a main storage device (memory) 33, an auxiliary storage device (memory) 35, a network interface 37, and a device interface 39. The information processing device 1 may be realized as the computer 30 in which the processor 31, the main storage device 33, the auxiliary storage device 35, the network interface 37, and the device interface 39 are connected via a bus 41.

[0017] Although the computer 30 shown in FIG. 2 includes one of each component, it may also include multiple of the same component. Furthermore, while FIG. 1 shows a single computer 30, the software may be installed on multiple computers, each of which may execute the same or different parts of the software. In this case, a distributed computing configuration may be used in which each computer communicates via a network interface 37 or the like to execute the processing. In other words, the information processing device 1 in this embodiment may be configured as a system in which one or more computers execute instructions stored in one or more storage devices to realize various functions described below. Furthermore, information transmitted from a terminal may be processed by one or more computers on a cloud, and the processing results may be transmitted to a terminal such as a display device (display unit) corresponding to the external device 9B.

[0018] Various computations of the information processing device 1 in this embodiment may be executed in parallel using one or more processors, or using multiple computers via a network. Furthermore, various computations may be distributed to multiple processor cores within a processor and executed in parallel. Furthermore, some or all of the processes, means, etc. disclosed herein may be executed by at least one of a processor and a storage device provided on a cloud that can communicate with the computer 30 via a network. Thus, various functions described below in this embodiment may be implemented in the form of parallel computing using one or more computers.

[0019] The processor 31 may be an electronic circuit (such as a processing circuit, processing circuitry, CPU (Central Processing Unit), GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), or ASIC (Application Specific Integrated Circuit)) including a control device and an arithmetic device of the computer 30. The processor 31 may also be a semiconductor device including a dedicated processing circuit. The processor 31 is not limited to an electronic circuit using electronic logic elements, and may be realized by an optical circuit using optical logic elements. The processor 31 may also include an arithmetic function based on quantum computing.

[0020] The processor 31 performs arithmetic processing based on data and software (programs) input from each device, etc., configured internally of the computer 30, and can output the arithmetic results and control signals to each device, etc. The processor 31 may control each component constituting the computer 30 by executing the OS (Operating System) of the computer 30, applications, etc.

[0021] The information processing device 1 in this embodiment may be realized by one or more processors 31. Here, the processor 31 may refer to one or more electronic circuits arranged on one chip, or may refer to one or more electronic circuits arranged on two or more chips or two or more devices. When multiple electronic circuits are used, the electronic circuits may communicate with each other via wire or wirelessly.

[0022] The main memory device 33 is a memory device that stores instructions executed by the processor 31 and various data, and information stored in the main memory device 33 is read by the processor 31. The auxiliary memory device 35 is a memory device other than the main memory device 33. Note that these memory devices refer to any electronic component capable of storing electronic information, and may be semiconductor memory. The semiconductor memory may be either volatile memory or non-volatile memory. The memory device for saving various data used in the information processing device 3 according to this embodiment may be realized by the main memory device 33 or the auxiliary memory device 35, or may be realized by an internal memory built into the processor 31. For example, the memory unit in this embodiment may be realized by the main memory device 33 or the auxiliary memory device 35.

[0023] Multiple processors may be connected (coupled) to one storage device (memory), or a single processor 31 may be connected. Multiple storage devices (memories) may be connected (coupled) to one processor. When the information processing device 1 in this embodiment is configured with at least one storage device (memory) and multiple processors connected (coupled) to this at least one storage device (memory), it may include a configuration in which at least one of the multiple processors is connected (coupled) to at least one storage device (memory). This configuration may also be realized by storage devices (memories) and processors 31 included in multiple computers. Furthermore, it may include a configuration in which the storage device (memory) is integrated with the processor 31 (for example, a cache memory including an L1 cache and an L2 cache).

[0024] The network interface 37 is an interface for connecting to the communication network 5 wirelessly or via a wire. The network interface 37 may be an appropriate interface, such as one that conforms to an existing communication standard. The network interface 37 may exchange information with an external device 9A connected via the communication network 5. The communication network 5 may be any one of a wide area network (WAN), a local area network (LAN), a personal area network (PAN), or a combination thereof, as long as information is exchanged between the computer 30 and the external device 9A. An example of a WAN is the Internet, an example of a LAN is IEEE 802.11 or Ethernet (registered trademark), and an example of a PAN is Bluetooth (registered trademark) or NFC (Near Field Communication), etc.

[0025] The device interface 39 is an interface such as a USB (Universal Serial Bus) that directly connects to an output device such as a display device, an input device, and the external device 9 B. The output device may also have a speaker that outputs sound and the like.

[0026] The external device 9A is a device connected to the computer 30 via a network, while the external device 9B is a device directly connected to the computer 30.

[0027] The external device 9A or the external device 9B may be, for example, an input device (input unit). The input device is, for example, a device such as a camera, a microphone, a motion capture device, various sensors, a keyboard, a mouse, or a touch panel, and provides acquired information to the computer 30. The external device 9A or the external device 9B may also be a device equipped with an input unit, a memory, and a processor, such as a personal computer, a tablet terminal, or a smartphone.

[0028] Furthermore, the external device 9A or the external device 9B may be, for example, an output device (output unit). The output device may be, for example, a display device (display unit) such as an LCD (Liquid Crystal Display), a CRT (Cathode Ray Tube), a PDP (Plasma Display Panel), or an organic EL (Electro Luminescence) panel, or may be a speaker that outputs sound, etc. Furthermore, the external device 9A or the external device 9B may be a device that includes an output device, a memory, and a processor, such as a personal computer, a tablet terminal, or a smartphone.

[0029] The external device 9A or the external device 9B may be a storage device (memory). For example, the external device 9A may be a network storage or the like, and the external device 9B may be a storage device such as an HDD.

[0030] Furthermore, the external device 9A or the external device 9B may be a device having some of the functions of the components of the information processing device 1 in this embodiment. In other words, the computer 30 may transmit or receive some or all of the processing results of the external device 9A or the external device 9B.

[0031] 3 is a diagram showing an example of functional blocks in the processor 31. The processor 31 may include, for example, a missing value interpolation unit 310 and a primary task analysis unit 312 as functions for realizing missing value interpolation processing and primary task processing. The functions realized by the missing value interpolation unit 310 and the primary task analysis unit 312 may be stored as programs in, for example, the main storage device 33 or the auxiliary storage device 35. The processor 31 can realize the functions related to the missing value interpolation unit 310 and the primary task analysis unit 312 by reading and executing the programs stored in the main storage device 33 or the auxiliary storage device 35.

[0032] When there is missing data in a detection data group obtained by a detector or the like (i.e., an uninterpolated analysis target data group), the missing value interpolation unit 310 calculates an interpolated value that interpolates the value of the missing data (missing value) based on a combination of feature quantities obtained using the uninterpolated analysis target data group, and uses this to generate an interpolated analysis target data group. Specifically, the missing value interpolation unit 310 includes an interpolation fixed processing unit 310A and a missing value calculation unit 310B.

[0033] The interpolation fixed processing unit 310A calculates a combination of feature quantities related to a detection data group obtained by a detector or the like. Here, a feature quantity related to the detection data group (hereinafter simply referred to as a "feature quantity") is a unique value obtained using the detection data group. A combination of feature quantities related to the detection data group (hereinafter simply referred to as a "combination of feature quantities") refers to a combination of different types of feature quantities.

[0034] Examples of feature quantities include the average value, maximum value, minimum value, median value, values ​​at points dividing the data range into four, and modes (these and values ​​subjected to statistical processing are collectively referred to as statistical values) obtained from the detection data group. Feature quantities are not limited to scalars, and may be expressed by vectors or tensors. A combination of multiple feature quantities may be any combination of scalars, vectors, and tensors.

[0035] For example, in a learning process described below, the interpolation fixed processing unit 310A calculates a combination of feature quantities for a plurality of patterns from a group of detection data (i.e., a group of non-interpolated analysis target data) obtained by a detector, etc. Furthermore, in an inference process described below, the interpolation fixed processing unit 310A may calculate a combination of feature quantities for a predetermined pattern from a group of detection data obtained by a detector, etc.

[0036] The missing value calculation unit 310B may input the combination of feature quantities calculated by the interpolation fixed processing unit 310A, calculate and output an interpolated value that interpolates the missing value. The missing value calculation unit 310B may be realized, for example, by a trained model as a missing value interpolation model. The trained model may be, for example, a deep neural network (DNN) model, a multi-layer perceptron (MLP) model, or the like.

[0037] The missing value calculation unit 310B may use the interpolated value to interpolate missing data in the non-interpolated analysis target data group to generate an interpolated analysis target data group. The missing value calculation unit 310B outputs the interpolated analysis target data group to the primary task analysis unit 312.

[0038] Furthermore, the missing value calculation unit 310B may adjust the network parameters as a missing value interpolation model according to the information from the evaluation unit 312B.

[0039] The calculation of the feature combination by the interpolation fixed processing unit 310A and the calculation of the interpolated value by the missing value calculation unit 310B may be alternately repeated. For example, the interpolation fixed processing unit 310A calculates the feature combination using a detection data group interpolated using the interpolated value calculated by the missing value calculation unit 310B. The missing value calculation unit 310B may input the feature combination based on the interpolated detection data group, calculate and output an interpolated value that interpolates the missing value. After these calculations are repeated as necessary, the missing value calculation unit 310B may generate an interpolated analysis target data group and output it to the primary task analysis unit 312.

[0040] The primary task analysis unit 312 may acquire the interpolated analysis target data group from the missing value interpolation unit 310 and execute evaluation processing in the primary task processing and learning processing. Specifically, the primary task analysis unit 312 may include a primary task calculation unit 312A and an evaluation unit 312B.

[0041] The primary task analysis unit 312 may input the interpolated analysis target data group output from the missing value interpolation unit 310, execute a predetermined primary task process (compression process), and output a calculated value (compressed value).

[0042] The primary task analysis unit 312 can be realized using, for example, a DNN model, an MLP model, or the like.

[0043] Furthermore, the primary task analysis unit 312 may adjust the network parameters as the primary task model according to the information from the evaluation unit 312B.

[0044] The evaluation unit 312B may operate in the learning process of the missing value calculation unit 310B or the like as a missing value interpolation model and may perform evaluation processing. That is, the evaluation unit 312B may acquire compressed values ​​from the primary task analysis unit 312 and evaluate whether the compressed values ​​are appropriate (e.g., to what extent the interpolated analysis target data group can be reproduced from the compressed values) using an objective function (evaluation processing). Here, for example, a loss function for DNN optimization may be used as the objective function. Specific examples of the objective function include, for example, a rate-distortion loss that represents the trade-off between the compression rate and the restoration error upon decompression if the primary task is data compression, and a cross-entropy loss that evaluates the recognition error rate if the primary task is image recognition.

[0045] More specifically, the evaluation unit 312B calculates the difference between the value of the objective function using the primary task calculated value and a threshold. If the calculated difference is not equal to or less than a reference value, the evaluation unit 312B may determine that the primary task calculated value is invalid. If the evaluation unit 312B determines that the primary task calculated value is invalid, the evaluation unit 312B may calculate learning feedback data for minimizing (or maximizing) the value of the objective function.

[0046] Here, the learning feedback data is information (such as gradient information of the DNN) for adjusting the network parameters of the missing value calculation unit 310B, or the missing value calculation unit 310B and the primary task analysis unit 312.

[0047] (Learning Process of Missing Value Interpolation Model) Next, a learning process of the missing value interpolation model realized by the information processing device 1 according to the embodiment will be described.

[0048] 5 is a flowchart showing an example of the flow of the learning process of the missing value interpolation model. The learning process of the missing value interpolation model will be described with reference to FIGS.

[0049] First, the interpolation fixed processing unit 310A may acquire a group of non-interpolated analysis target data (detection data group) from a detector or the like (step S1).

[0050] The interpolation fixed processing unit 310A may calculate a combination of feature quantities of a specific pattern using the acquired group of non-interpolated analysis target data (step S2).

[0051] The missing value calculation unit 310B may input the combination of feature amounts calculated by the interpolation fixed processing unit 310A, and calculate and output an interpolation value that interpolates the missing value (step S3).

[0052] The missing value calculation unit 310B may generate an interpolated analysis target data group using the interpolated value and output it to the primary task analysis unit 312 (step S4).

[0053] The primary task analysis unit 312 may input the interpolated analysis target data group output from the missing value interpolation unit 310, execute a predetermined primary task process (data compression process), and output the obtained value of the objective function (step S5).

[0054] The evaluation unit 312B may obtain the value of the objective function from the primary task analysis unit 312 and calculate the value of the objective function (step S6).

[0055] The evaluation unit 312B may determine whether the primary task calculation value is appropriate based on the value of the objective function (step S7). For example, the evaluation unit 312B may determine whether the primary task calculation value is appropriate based on whether the value of the objective function converges within a predetermined range. If the evaluation unit 312B determines that the primary task calculation value is not appropriate (No in step S7), the evaluation unit 312B may generate learning feedback data for minimizing (or maximizing) the value of the objective function and output the learning feedback data to the missing value calculation unit 310B and the primary task analysis unit 312 (step S8).

[0056] The missing value calculation unit 310B and the primary task analysis unit 312 may adjust the network parameters based on the learning feedback data (step S9).

[0057] On the other hand, if it is determined that the primary task calculation value is appropriate (Yes in step S7), the evaluation unit 312B may end the learning process.

[0058] The processes from step S1 to step S7 may be repeated until it is determined in step S7 that the main task calculation value is valid.

[0059] The learning process may be performed for each combination of feature quantities. The learning process may also be performed using multiple groups of uninterpolated analysis target data. For example, in step S7, the user may select the combination of feature quantities that has the smallest difference between the objective function value and the reference value as the combination of feature quantities to be used in the missing value interpolation process, and use the selected combination of feature quantities as input. The missing value calculation unit 310B, in which the network parameters have been optimized, may then be employed as the trained missing value interpolation model. Further optimization of the network parameters may be performed for the selected combination of feature quantities according to the learning process.

[0060] (Inference Processing: Missing Value Interpolation Processing) Next, a missing value interpolation processing as an inference processing realized by the information processing device 1 according to the embodiment will be described.

[0061] 4 is a diagram for explaining the flow of data between functional blocks when performing inference processing. In FIG. 4, it is assumed that the network parameters of the missing value interpolation model of the missing value calculation unit 310B and the primary task model of the primary task analysis unit 312 are optimized and fixed by the learning process shown in FIG.

[0062] 6 is a flowchart showing an example of the flow of the inference process executed by the information processing device 1. The inference process executed by the information processing device 1 will be described with reference to FIGS.

[0063] First, the interpolation fixed processing unit 310A may acquire a group of non-interpolated analysis target data (detection data group) from a detector or the like (step S1).

[0064] The interpolation fixed processing unit 310A may use the acquired group of non-interpolated analysis target data to calculate a combination of feature amounts of the pattern determined by the learning process (step S2).

[0065] The missing value calculation unit 310B may input the combination of feature amounts calculated by the interpolation fixed processing unit 310A, and calculate and output an interpolation value that interpolates the missing value (step S3).

[0066] The missing value calculation unit 310B may generate an interpolated analysis target data group using the interpolated value and output it to the primary task analysis unit 312 (step S4).

[0067] The primary task analysis unit 312 may input the interpolated analysis target data group output from the missing value interpolation unit 310, execute a predetermined primary task process (data compression process), and output the obtained compressed value (step S5).

[0068] The primary task analysis unit 312 may output the obtained compressed value to an external device via a network (step S9).

[0069] The above process may be repeatedly executed each time a group of uninterpolated data to be analyzed is acquired from a detector or the like.

[0070] The information processing device 1 according to the embodiment may include a main storage device 33 and an auxiliary storage device 35 as at least one memory, and a processor 31 as at least one processor. The information processing device 1 is implemented by the processor 31. The missing value interpolation unit 310 implemented by the processor 31 may input multiple feature quantities calculated from a data group output from a detection device and having at least one missing value (an uninterpolated analysis target data group), and output at least one interpolated value. The missing value interpolation unit 310 may generate an interpolated analysis target data group as a second data group using the uninterpolated analysis target data group and at least one interpolated value. The primary task analysis unit 312 implemented by the processor 31 inputs the interpolated analysis target data group and outputs a predetermined inference value, such as a compressed value.

[0071] That is, the information processing device 1 according to the embodiment may obtain at least one interpolated value from multiple feature quantities calculated from a group of uninterpolated data to be analyzed, and generate an interpolated group of data to be analyzed by performing missing value interpolation using the interpolated value. Therefore, compared to when missing value interpolation is performed using a fixed value such as an average value, missing value interpolation according to data characteristics can be achieved. As a result, the deviation (deterioration) of data characteristics before and after missing value interpolation can be suppressed compared to conventional methods, enabling more optimized missing value interpolation processing. Furthermore, compared to when multiple regression or tensor decomposition is performed, missing value interpolation can be achieved with a smaller computational load.

[0072] According to the information processing device 1 of the embodiment, by inputting a plurality of feature quantities calculated from a group of non-interpolated analysis target data, it is possible to obtain the interpolated values ​​required for interpolation. Therefore, it is possible to quickly and easily interpolate missing values ​​with a smaller calculation load than conventional methods. This is particularly beneficial when missing or invalid values ​​are randomly included in the detection data group output from the detection device, and when video shooting or continuous observation is required for real-time processing.

[0073] Furthermore, the processor 31 may execute a learning process for the missing value interpolation model included in the missing value interpolation unit 310 by using the inferred value and a predetermined evaluation function through the main task processing.

[0074] Once a missing value interpolation model is constructed through the above learning process, an interpolated analysis target data group can be generated by performing a forward propagation inference calculation using an uninterpolated analysis target data group as input. Therefore, missing value interpolation can be achieved quickly and easily with a smaller computational load than conventional methods.

[0075] The processor 31 can also execute a learning process for the main task model.

[0076] Therefore, the learning of the main task model and the learning of the missing value interpolation model can be performed simultaneously, which results in the optimization of the main task model and the missing value interpolation model at low cost.

[0077] (Variation 1) In the above embodiment, the learning process is exemplified as a learning process in which the network parameters of both the missing value calculation unit 310B and the primary task analysis unit 312 are adjusted based on learning feedback data. However, the learning processes of the missing value calculation unit 310B and the primary task analysis unit 312 may be performed separately. Furthermore, the learning process that adjusts the network parameters of both the missing value calculation unit 310B and the primary task analysis unit 312 may be combined with independent learning processes that adjust the network parameters of the missing value calculation unit 310B and the primary task analysis unit 312 separately.

[0078] (Variation 2) In the above embodiment, a case where one interpolated value is calculated in the missing value interpolation process has been described as an example. However, this is not limiting, and multiple interpolated values ​​may be calculated in the missing value interpolation process. For example, the user may interpolate missing data using a secondary interpolated value obtained using the multiple calculated interpolated values, depending on the characteristics of the uninterpolated analysis target data.

[0079] (Modification 3) In the above embodiment, the feature amounts are expressed using scalars, vectors, tensors, etc. However, it is also possible to use feature amounts expressed as histograms.

[0080] For example, the missing value interpolation process according to the embodiment can be realized by installing a program for executing the missing value interpolation process in a computer provided in the detection device or in a computer provided along a route or at a relay point in a network connecting the detection device and a predetermined storage device, and then expanding the program in memory. In this case, the program that can cause a computer to execute the method can also be stored and distributed on a storage medium such as a magnetic disk (such as a hard disk), an optical disk (such as a CD-ROM or DVD), or a semiconductor memory. The procedure and effects of the optimal parameter determination process using the determination program are the same as those in the embodiment, and therefore will not be described here.

[0081] Some or all of the devices in the above-described embodiments may be configured as hardware, or may be configured as information processing software (programs) executed by a CPU, GPU, or the like. In the case of software information processing, software that realizes at least some of the functions of each device in the above-described embodiments may be stored on a non-transitory storage medium (non-transitory computer-readable medium) such as a flexible disk, CD-ROM (Compact Disc-Read Only Memory), or USB memory, and the software information processing may be executed by loading the software into the computer 30. The software may also be downloaded via the communication network 5. Furthermore, the software may be implemented in a circuit such as an ASIC or FPGA, so that the information processing is executed by hardware.

[0082] The type of storage medium that stores the software is not limited. The storage medium is not limited to removable media such as magnetic disks or optical disks, but may be fixed storage media such as hard disks or memory. The storage medium may be provided inside the computer or outside the computer.

[0083] In this specification (including the claims), when the expression "at least one of a, b, and c" or "at least one of a, b, or c" (including similar expressions) is used, it includes any of a, b, c, a-b, a-c, bc, or a-bc. It may also include multiple instances of any element, such as a-a, a-bb-b, a-a-bb-cc-c, etc. Furthermore, it also includes the addition of elements other than the listed elements (a, b, and c), such as having d, as in a-bc-d.

[0084] In this specification (including the claims), when expressions such as "using data as input / based on / according to / in response to" (including similar expressions) are used, unless otherwise specified, this includes cases where various data itself is used as input, or where various data that has been processed in some way (e.g., noise-added, normalized, intermediate representation of various data, etc.) is used as input. Furthermore, when it is stated that a result is obtained "based on / according to / in response to data," this includes cases where the result is obtained based solely on the data in question, as well as cases where the result is obtained in response to other data, factors, conditions, and / or states other than the data in question. Furthermore, when it is stated that "data is output," unless otherwise specified, this includes cases where various data itself is used as output, or where various data that has been processed in some way (e.g., noise-added, normalized, intermediate representation of various data, etc.) is output.

[0085] When the terms "connected" and "coupled" are used in this specification (including the claims), they are intended as open-ended terms that include any of direct connection / coupling, indirect connection / coupling, electrically connection / coupling, communicatively connection / coupling, functionally connection / coupling, and physically connection / coupling. These terms should be interpreted appropriately depending on the context in which they are used, but any connection / coupling form that is not intentionally or naturally excluded should be interpreted as being included in these terms without any restriction.

[0086] In this specification (including the claims), when the expression "A configured to B" is used, it may include a situation in which the physical structure of element A has a configuration capable of performing operation B, and a permanent or temporary setting / configuration of element A is configured / set to actually perform operation B. For example, when element A is a general-purpose processor, it is sufficient that the processor has a hardware configuration capable of performing operation B, and is configured to actually perform operation B by setting a permanent or temporary program (instruction). Furthermore, if element A is a dedicated processor or dedicated arithmetic circuit, etc., it is sufficient that the circuit structure of the processor is implemented to actually execute operation B, regardless of whether control instructions and data are actually attached.

[0087] When used in this specification (including the claims), terms implying containing or possessing (e.g., "comprising," "including," "having," etc.) are intended to be open-ended terms that include cases in which something other than the object indicated by the object of the term is contained or possessed. When the object of such a term implies no quantity or a singular number (e.g., an article such as "a" or "an"), the expression should be construed as not being limited to a specific number.

[0088] In this specification (including the claims), although expressions such as "one or more" or "at least one" are used in some places and expressions that do not specify a quantity or that imply a singular number (expressions using the articles "a" or "an") are used in other places, the latter expressions are not intended to mean "one." In general, expressions that do not specify a quantity or that imply a singular number (expressions using the articles "a" or "an") should be interpreted as not necessarily being limited to a specific number.

[0089] In this specification, when a particular advantage / result is described as being obtained with respect to a particular configuration of a certain embodiment, it should be understood that the same advantage / result can also be obtained with one or more other embodiments having the same configuration, unless otherwise stated. However, it should be understood that the presence or absence of the effect generally depends on various factors, conditions, and / or states, etc., and that the effect is not necessarily obtained with the configuration. The effect is merely obtained by the configuration described in the embodiment when various factors, conditions, and / or states, etc. are satisfied, and the effect does not necessarily occur in a claimed invention that defines the same configuration or a similar configuration.

[0090] When terms such as "maximize" are used in this specification (including the claims), they include finding a global maximum, finding an approximation of a global maximum, finding a local maximum, and finding an approximation of a local maximum, and should be interpreted appropriately according to the context in which the term is used. They also include probabilistic or heuristic approximations of these maxima. Similarly, when terms such as "minimize" are used, they include finding a global minimum, finding an approximation of a global minimum, finding a local minimum, and finding an approximation of a local minimum, and should be interpreted appropriately according to the context in which the term is used. They also include probabilistic or heuristic approximations of these minima. Similarly, when terms such as "optimize" are used, they include finding a global optimum, finding an approximation of a global optimum, finding a local optimum, and finding an approximation of a local optimum, and should be interpreted appropriately according to the context in which the term is used. It also includes finding approximations of these optimum values ​​probabilistically or heuristically.

[0091] In this specification (including claims), when multiple pieces of hardware perform a predetermined process, the pieces of hardware may cooperate to perform the predetermined process, or some of the hardware may perform all of the predetermined process. Furthermore, some of the hardware may perform part of the predetermined process, and other hardware may perform the rest of the predetermined process. In this specification (including claims), when an expression such as "one or more pieces of hardware perform a first process, and the one or more pieces of hardware perform a second process" is used, the hardware performing the first process and the hardware performing the second process may be the same or different. In other words, it is sufficient that the hardware performing the first process and the hardware performing the second process are included in the one or more pieces of hardware. Note that the hardware may include an electronic circuit or a device including an electronic circuit.

[0092] In this specification (including the claims), when multiple storage devices (memories) store data, each of the multiple storage devices (memories) may store only a portion of the data, or may store the entire data.

[0093] Although the embodiments of the present disclosure have been described in detail above, the present disclosure is not limited to the individual embodiments described above. Various additions, modifications, substitutions, partial deletions, etc. are possible within the scope of the conceptual idea and spirit of the present invention derived from the content defined in the claims and their equivalents. For example, in all of the above-described embodiments, when numerical values ​​or formulas are used in the explanation, they are shown as examples and are not limited to these. Furthermore, the order of each operation in the embodiments is shown as an example and is not limited to these.

[0094] REFERENCE SIGNS LIST 1 Information processing device 5 Communication network 9A External device 9B External device 30 Computer 31 Processor 33 Main memory device 35 Auxiliary memory device 37 Network interface 39 Device interface 41 Bus 310 Missing value interpolation unit 310A Fixed processing unit for interpolation 310B Missing value calculation unit 312 Primary task analysis unit 312A Primary task calculation unit 312B Evaluation unit

Claims

1. An information processing device comprising at least one memory and at least one processor, wherein the at least one processor inputs a plurality of feature amounts calculated from a first data group output from a detection device and having at least one missing value into a first model, and outputs at least one interpolated value used to interpolate the first data group having at least one missing value.

2. The information processing device according to claim 1, wherein the at least one processor inputs a second set of data generated using the first set of data and the at least one interpolated value into a second model, and outputs an inferred value generated using the second set of data.

3. An information processing device comprising at least one memory and at least one processor, wherein the at least one processor inputs a plurality of feature quantities calculated from a first data group output from a detection device and having at least one missing value into a first model, outputs at least one interpolated value used to interpolate the first data group having at least one missing value, inputs a second data group generated using the first data group and the at least one interpolated value into a second model, outputs an inferred value generated using the second data group, and executes a learning process of the first model using the inferred value and a predetermined evaluation function, in which the plurality of feature quantities are input and at least one interpolated value is output.

4. The information processing device according to claim 3, wherein the at least one processor executes a learning process for a second model that inputs the second data group and outputs the inferred value using the inferred value and the predetermined evaluation function.

5. An information processing device according to any one of claims 1 to 4, wherein the plurality of feature amounts include at least one of an average value, a maximum value, a minimum value, a median value, a value of a point dividing a data range into four, and a mode value.

6. The information processing device according to claim 1 , wherein the plurality of feature quantities include at least one of a scalar quantity, a vector quantity, and a tensor quantity.

7. An information processing device according to any one of claims 2 to 4, wherein the at least one processor calculates new features from the second data group, inputs the new features, outputs at least one new interpolated value, and generates the second data group using the first data group and the at least one new interpolated value.

8. The information processing device according to claim 2, wherein the inferred value is a value obtained by compressing the second data group.

9. An information processing device according to any one of claims 1 to 4, wherein the plurality of feature quantities include at least two or more different feature quantities among an average value, a maximum value, a minimum value, a median value, a value of a point dividing a data range into four, and a mode value.

10. The information processing device according to claim 1, wherein the plurality of feature amounts are statistical values.

11. An information processing program that causes a computer to input a plurality of feature amounts calculated from a data group output from a detection device and having at least one missing value into a first model, and output at least one interpolated value used to interpolate the data group having at least one missing value.

12. An information processing method comprising: inputting a plurality of feature quantities calculated from a data group output from a detection device and having at least one missing value into a first model; and outputting at least one interpolated value used to interpolate the data group having at least one missing value.

Citation Information

Patent Citations

  • Recognition device and deficiency value estimating and learning method

    JP1996212184A

  • Data processing apparatus, data processing method, and program

    JP2020052886A