WO2023073910A1 - モデル分析装置、モデル分析方法、及び記録媒体 - Google Patents

モデル分析装置、モデル分析方法、及び記録媒体 Download PDF

Info

Publication number
WO2023073910A1
WO2023073910A1 PCT/JP2021/039981 JP2021039981W WO2023073910A1 WO 2023073910 A1 WO2023073910 A1 WO 2023073910A1 JP 2021039981 W JP2021039981 W JP 2021039981W WO 2023073910 A1 WO2023073910 A1 WO 2023073910A1
Authority
WO
WIPO (PCT)
Prior art keywords
configuration information
model
verification target
target
verification
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/JP2021/039981
Other languages
English (en)
French (fr)
Japanese (ja)
Inventor
淳 西岡
純明 榮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NEC Corp
Original Assignee
NEC Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by NEC Corp filed Critical NEC Corp
Priority to PCT/JP2021/039981 priority Critical patent/WO2023073910A1/ja
Priority to US18/694,032 priority patent/US20240385943A1/en
Priority to JP2023556017A priority patent/JP7666628B2/ja
Publication of WO2023073910A1 publication Critical patent/WO2023073910A1/ja
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00—Error detection; Error correction; Monitoring
    • G06F11/36—Prevention of errors by analysis, debugging or testing of software
    • G06F11/3604—Analysis of software for verifying properties of programs
    • G06F11/3608—Analysis of software for verifying properties of programs using formal methods, e.g. model checking, abstract interpretation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/445—Program loading or initiating
    • G06F9/44505—Configuring for program initiating, e.g. using registry, configuration files
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00—Error detection; Error correction; Monitoring
    • G06F11/36—Prevention of errors by analysis, debugging or testing of software
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Definitions

  • the present disclosure relates to a model analysis device, a model analysis method, and a recording medium.
  • Patent Literature 1 discloses a server device that can select and supply optimal trained models for various devices with different environments and conditions.
  • Patent Document 1 simply selects a model that fits between different devices based on the data used to generate the trained model. In operation verification of network equipment, it is necessary to select a model considering the software information that configures the equipment. However, there are various combinations of software used in equipment, and it takes time and effort to select a suitable model.
  • An example of the purpose of the present disclosure is to provide a model analysis device that can easily search for suitable trained models in operation verification of network devices.
  • a model analysis apparatus includes target information acquisition means for acquiring device configuration information including hardware configuration information for a verification target and software configuration information including firmware configuration information and protocol processing software configuration information. , model information acquisition means for acquiring device configuration information in the environment in which the trained model used for verification of the verification target was generated; a suitability evaluation means for evaluating suitability with respect to the finished model; and an output means for outputting the evaluated result.
  • a model analysis apparatus includes target information acquisition means for acquiring device configuration information including hardware configuration information for a verification target and software configuration information including firmware configuration information and protocol processing software configuration information. , model retrieval means for retrieving a trained model suitable for verification of a verification target based on the obtained device configuration information; and output means for outputting the retrieved trained model.
  • a model analysis method acquires device configuration information including hardware configuration information in a verification target and software configuration information including firmware configuration information and protocol processing software configuration information, and verifies a verification target. Acquire the device configuration information in the environment where the trained model used for the verification was generated, and evaluate the suitability for the trained model to be verified based on the obtained device configuration information for the verification target and the trained model. output the results.
  • Another model analysis method acquires device configuration information including hardware configuration information in a verification target and software configuration information including firmware configuration information and protocol processing software configuration information, Based on the obtained device configuration information, it searches for a trained model that matches the verification target, and outputs the searched trained model.
  • a recording medium acquires device configuration information including hardware configuration information to be verified and software configuration information including firmware configuration information and protocol processing software configuration information, and performs verification of the verification target. Acquire the configuration information of the device in the environment where the trained model to be used was generated, evaluate the compatibility with the trained model to be verified based on the acquired device configuration information of the verification target and the trained model, and evaluate stores a program that causes a computer to output the results of
  • Another recording medium acquires device configuration information including hardware configuration information to be verified and software configuration information including firmware configuration information and protocol processing software configuration information, and acquires Based on the configuration information of the device, a trained model that matches the verification target is searched, and the searched trained model is output.
  • One example of the effects of the present disclosure is that it is possible to provide a model analysis device that can easily search for suitable trained models in the operation verification of network devices.
  • FIG. 1 is a block diagram showing the configuration of the model analysis device according to the first embodiment.
  • FIG. 2 is a diagram showing a hardware configuration in which the model analysis device according to the first embodiment is realized by a computer device and its peripheral devices.
  • FIG. 3 is a table showing a storage example of device configuration information in the first embodiment.
  • FIG. 4 is a flow chart showing the operation of model analysis in the first embodiment.
  • FIG. 5 is a block diagram showing the configuration of a model analysis device in a modification of the first embodiment.
  • FIG. 6 is a block diagram showing the configuration of the model analysis device in the second embodiment.
  • FIG. 7 is an example of a display screen of the compatible model search tool in the second embodiment.
  • FIG. 8 is another example of the display screen of the compatible model search tool in the second embodiment.
  • FIG. 9 is a flowchart showing model analysis operations in the second embodiment.
  • FIG. 1 is a block diagram showing the configuration of a model analysis device 100 according to the first embodiment.
  • the model analysis device 100 uses a learned model generated in its own environment in the past and other It is a device for verifying whether a trained model generated in the business operator's environment is suitable.
  • the model analysis device 100 includes a target information acquisition unit 101, a model information acquisition unit 102, a suitability evaluation unit 103, and an output unit 104.
  • FIG. The model analysis device 100 which is an essential component of this embodiment, will be described in detail below.
  • FIG. 2 is a diagram showing an example of a hardware configuration in which the model analysis device 100 according to the first embodiment of the present disclosure is realized by a computer device 500 including a processor.
  • the model analysis device 100 includes memory such as a CPU (Central Processing Unit) 501, ROM (Read Only Memory) 502, RAM (Random Access Memory) 503, and storage such as a hard disk for storing the program 504. It includes a device 505, a communication I/F (Interface) 508 for network connection, and an input/output interface 511 for inputting/outputting data.
  • the configuration information acquired by the target information acquisition unit 101 and the model information acquisition unit 102 is input to the model analysis device 100 via the input/output interface 511, for example.
  • the CPU 501 operates the operating system and controls the entire model analysis apparatus 100 according to the first embodiment of the present invention. Also, the CPU 501 reads programs and data from a recording medium 506 mounted in a drive device 507 or the like to a memory. Further, the CPU 501 functions as the target information acquisition unit 101, the model information acquisition unit 102, the suitability evaluation unit 103, the output unit 104, and a part of these in the first embodiment, and executes the program shown in FIG. Execute the processing or instructions in the flow chart shown in FIG.
  • the recording medium 506 is, for example, an optical disk, a flexible disk, a magneto-optical disk, an external hard disk, or a semiconductor memory.
  • a part of the recording medium of the storage device is a non-volatile storage device, in which programs are recorded.
  • the program may be downloaded from an external computer (not shown) connected to a communication network.
  • the input device 509 is realized by, for example, a mouse, keyboard, built-in key buttons, etc., and is used for input operations.
  • the input device 509 is not limited to a mouse, keyboard, or built-in key buttons, and may be a touch panel, for example.
  • the output device 510 is implemented by, for example, a display and used to confirm the output.
  • the first embodiment shown in FIG. 1 is implemented by the computer hardware shown in FIG.
  • the implementation means of each unit included in the model analysis apparatus 100 of FIG. 1 is not limited to the configuration described above.
  • the model analysis apparatus 100 may be implemented by one physically coupled device, or may be implemented by two or more physically separated devices connected by wire or wirelessly. good.
  • input device 509 and output device 510 may be connected to computer device 500 via a network.
  • the model analysis device 100 in the first embodiment shown in FIG. 1 can also be configured by cloud computing or the like.
  • the target information acquisition unit 101 is means for acquiring device configuration information including hardware configuration information and software configuration information consisting of firmware configuration information and protocol processing software configuration information in the verification target.
  • Verification in this embodiment means verifying whether anomalies such as operation and security of network device hardware and software (firmware and protocol processing software) are included.
  • Verification targets also include a configuration consisting of a plurality of network devices.
  • a hardware configuration is, for example, chip configuration information that controls the operation of a device.
  • the chip configuration information includes chip architecture, manufacturer name, model number, and the like.
  • Firmware information is information about software for operating hardware, and includes, for example, a firmware name and version number.
  • the protocol information includes protocol name and protocol processing software (routing software) information.
  • the protocol name indicates, for example, the protocol name of layer 3, which is the “network layer” in the third layer of the OSI (Open Systems Interconnection) reference model.
  • Specific examples of protocol names include BGP (Border Gateway Protocol), OSPF (Open Shortest Path First), STP (Spanning Tree Protocol), and the like.
  • Examples of protocol processing software include "GoBGP" or "FRR” if the protocol is BGP, and "FRR” or “Quagga” if the protocol is OSPF.
  • FIG. 3 is a table showing a storage example of device configuration information in this embodiment.
  • the model number of the chip name "ABC1111chip” is stored as the hardware configuration information.
  • the firmware configuration information the firmware name FSP (Intel Firmware Support Package) and its version "v0.9” are stored.
  • the protocol configuration information the protocol name "STP”, its protocol processing software "FRR” and version "v6.1” information are stored.
  • the target information acquisition unit 101 acquires the configuration information of the device to be verified, for example, triggered by an operation for verification by the user.
  • the target information acquisition unit 101 may search these pieces of information stored in the storage device 505 .
  • identifier information to be verified and configuration information of the device to be verified are stored in a linked state.
  • the target information acquisition unit 101 can acquire the configuration information associated with the ID stored in the storage device 505 by accepting input of identifier information such as an ID assigned to the verification target by the user.
  • the configuration information may be stored in another configuration (for example, configuration information storage means) instead of the storage device 505 .
  • the target information acquisition unit 101 outputs the acquired configuration information of the verification target to the compatibility evaluation unit 103 .
  • the model information acquisition unit 102 is means for acquiring device configuration information in the environment in which the trained model used for verification of the verification target is generated.
  • a learned model is a model generated by machine learning in order to output verification results based on verification data of normal and abnormal cases in the past in the same or similar configuration, for example.
  • Trained models include, but are not limited to, decision tree models, linear regression models, logistic regression models, neural networks models, and the like.
  • the model information acquisition unit 102 acquires device configuration information of a learned model used for verification of a verification target, for example, by receiving an input from the user through the input device 509 . Further, when the configuration information of the learned model device is stored in the storage device 505 , the model information acquisition unit 102 may acquire the configuration information from the storage device 505 .
  • the device configuration information here is configuration information that can be compared with the configuration information acquired by the target information acquisition unit 101 . That is, it is device configuration information including hardware configuration information and software configuration information including firmware configuration information and protocol processing software configuration information.
  • the model information acquisition unit 102 outputs the acquired configuration information of the learned model to the suitability evaluation unit 103 .
  • the suitability evaluation unit 103 is means for evaluating the suitability of a trained model to be verified based on the acquired configuration information of the device to be verified and the learned model.
  • the compatibility evaluation unit 103 evaluates the compatibility between the verification target and the learned model. Specifically, the suitability evaluation unit 103 compares each piece of configuration information and evaluates the presence or absence of suitability based on the degree of similarity of the configuration information.
  • the suitability evaluation unit 103 evaluates that there is suitability. On the other hand, if the similarity of the configuration information is not within the predetermined range, the suitability evaluation unit 103 evaluates that there is no suitability.
  • "Within a predetermined range in the hardware configuration information” means, for example, when chip names are compared as hardware configuration information, they are all from the same manufacturer and in the same series.
  • the suitability evaluation unit 103 determines the degree of similarity with respect to the hardware configuration information, taking into consideration OEM (Original Equipment Manufacturing) supply information. Specifically, if the manufacturer and product information match due to OEM supply, the conformity evaluation unit 103 determines that the chip name and manufacturer name are the same even if the chip name and manufacturer name are different.
  • the predetermined range in the software information is, for example, the case where the firmware name and protocol name are the same, and the only software version difference is a minor version. However, it is not limited to this range as long as the software operations can be regarded as substantially equivalent.
  • the suitability evaluation unit 103 determines the degree of similarity with respect to the software configuration information, taking into consideration the software compatibility information. For example, the suitability evaluation unit 103 considers that "FRR2.0" and "Quagga1.1" are compatible, and judges the degree of similarity assuming that these pieces of software information are the same. Suitability evaluation section 103 thus determines similarity and outputs suitability to output section 104 .
  • the output unit 104 is means for outputting the results evaluated by the suitability evaluation unit 103 .
  • the output unit 104 outputs the suitability evaluation result to the output device 510 or the like.
  • Adaptability evaluation is evaluation of whether or not the trained model to be applied is suitable under its own configuration environment.
  • model analysis device 100 configured as above will be described with reference to the flowchart of FIG.
  • FIG. 4 is a flow chart showing an overview of the operation of the model analysis device 100 in the first embodiment. Note that the processing according to this flowchart may be executed based on program control by the processor described above.
  • the target information acquisition unit 101 first acquires the configuration information of the device to be verified (step S101).
  • the model information acquisition unit 102 acquires the configuration information of the device in the environment in which the trained model used for verification of the verification target is generated (step S102).
  • the suitability evaluation unit 103 evaluates the suitability of the learned model to be verified based on the acquired configuration information of the device to be verified and the learned model (step S103).
  • the output unit 104 outputs the result of compatibility evaluated by the compatibility evaluation unit 103 (step S104). With this, the model analysis device 100 ends the model analysis operation.
  • the compatibility evaluation unit 103 analyzes the acquired hardware configuration information of the verification target and the learned model, software configuration information including firmware configuration information and protocol processing software configuration information, and Based on the configuration information of the device, including As a result, even in an operating environment that includes multiple pieces of software, searching for a suitable model does not take much time.
  • FIG. 5 is a block diagram showing the configuration of the model analysis device 110 in the modified example of the first embodiment.
  • a model analysis device 110 in the modification of the first embodiment includes a target information acquisition unit 111 , a model information acquisition unit 112 , a fitness calculation unit 113 , a fitness evaluation unit 114 and an output unit 115 .
  • the modified example of the first embodiment is significantly different in that it includes a goodness-of-fit calculation unit 113 in addition to the configuration of the first embodiment.
  • the modification of the first embodiment also differs from the first embodiment in that network configuration information is used as configuration information in addition to hardware configuration information and software configuration information.
  • the target information acquisition unit 111 acquires network configuration information in addition to verification target hardware configuration information and software configuration information.
  • Network configuration information is information that indicates connection relationships between hardware.
  • the notation "A ⁇ B ⁇ C ⁇ D" is used for network configuration information. This indicates that device A and device B are directly connected, device B and device C are directly connected, and device C and device D are directly connected.
  • the method by which the target information acquisition unit 111 acquires network configuration information is the same as the method by which the target information acquisition unit 101 acquires hardware information and software information.
  • the model information acquisition unit 112 acquires network configuration information in addition to the hardware configuration information and software information that generated the trained model used for verification of the verification target.
  • the method by which the model information acquisition unit 112 acquires network configuration information is the same as the method by which the model information acquisition unit 112 acquires hardware information and software information.
  • the fitness calculation unit 113 calculates the degree of fitness based on the similarity of the hardware configuration, software information, and network configuration information of the verification target and the trained model acquired by the target information acquisition unit 111 and the model information acquisition unit 112. Calculate the fitness shown.
  • the degree-of-fit calculation unit 113 sets a coefficient indicating the degree of similarity for each configuration.
  • the suitability calculation unit 113 sets "1.0". If the hardware configuration is the same manufacturer and the same series, it is set to "0.8". If the software configuration is the same software and the same version, the compatibility calculation unit 113 determines "1.0". Even if the software is the same, if the versions are different, the coefficient is changed according to the version difference. For example, if the different part is a minor version name difference, it is "0.9", and if it is a major version name difference, it is "0.5".
  • the compatibility calculation unit 113 also calculates the compatibility of the network configuration information based on the similarity.
  • the degree of similarity of network configuration information indicates the similarity of the connection relationship of devices among constituent devices. For example, when the network configuration of itself is "A ⁇ B ⁇ C ⁇ D", the compatibility calculation unit 113 determines that the network connection configuration of "A ⁇ B ⁇ C” is the same (for example, "D ⁇ A ⁇ B ⁇ C”), and the coefficient is set to “0.8”.
  • the fitness calculation unit 113 calculates the fitness by averaging the calculated coefficients.
  • the compatibility calculation unit 113 may calculate the compatibility using the network configuration information only when the hardware configuration and the software configuration completely match.
  • the method of calculating the degree of conformity by the degree of conformity calculation unit 113 is not limited to this as long as it is a method capable of calculating the degree of similarity of each configuration.
  • the suitability evaluation unit 114 evaluates suitability for the trained model to be verified based on the suitability calculated by the suitability calculation unit 113 .
  • the suitability evaluation unit 114 determines that there is suitability when the calculated degree of suitability is equal to or greater than the threshold.
  • the suitability evaluation unit 114 determines that there is no suitability when the calculated degree of suitability is smaller than the threshold.
  • the threshold is determined in advance and stored in the storage device 505, for example.
  • the output unit 115 outputs the results evaluated by the suitability evaluation unit 114.
  • the output unit 115 evaluates suitability as an evaluation of whether or not the trained model to be adapted is suitable under its own configuration environment.
  • the output unit 115 may output the fitness result calculated by the fitness calculation unit 113 .
  • the suitability evaluation unit 114 evaluates the suitability for the trained model to be verified based on the suitability calculated by the suitability calculation unit 113 . Accordingly, the suitability evaluation unit 114 determines that there is suitability if, for example, even if the hardware configuration information is different, the software configuration information is the same and the degree of suitability is equal to or greater than the threshold. Therefore, as a result of comprehensively judging the configuration information of the device, it is possible to evaluate whether or not the learned model is suitable.
  • the model analysis device 120 in the second embodiment is a device for retrieving which model is suitable for its own device configuration when learning models are shared among a plurality of business operators.
  • Each component in each embodiment of the present disclosure can of course be implemented in hardware, as in the computer device shown in FIG.
  • FIG. 6 is a block diagram showing the configuration of the model analysis device 120 according to the second embodiment of the present disclosure.
  • the model analysis device 120 according to the second embodiment will be described, focusing on the parts different from the model analysis device 100 according to the first embodiment.
  • a model analysis device 120 according to the second embodiment includes a target information acquisition unit 121 , a model search unit 122 , an output unit 123 and a result reception unit 124 .
  • the target information acquisition unit 121 By inputting identifier information such as an ID assigned to a device to be verified, the target information acquisition unit 121 in this embodiment can acquire configuration information to be verified linked to the identifier information.
  • the configuration information is device configuration information including hardware configuration information and software configuration information including firmware configuration information and protocol processing software configuration information.
  • the target information acquisition unit 121 outputs the acquired device configuration information to the model search unit 122 .
  • the model search unit 122 searches for a suitable learned model based on the configuration information of the device to be verified.
  • configuration information to be verified is input from the target information acquisition unit 121, it searches for a learned model that was generated in the past with the same or similar configuration by each business operator.
  • similarity refers to a case where the degree of similarity described in the first embodiment is within a predetermined range.
  • a trained model is a trained model generated by machine learning to output verification results based on verification data for normal and abnormal conditions in each business operator.
  • the output unit 123 is means for displaying and outputting the learned model retrieved by the model retrieval unit 122 on the output device 510 or the like.
  • the output unit 123 outputs, as metadata, the purpose of the model, the format of the data used in the verification data (time, tx_pkt_cnt, rx_pkt_cnt), supplementary information such as usage, or URL information indicating the storage location of the trained model. I don't mind.
  • FIG. 7 is an example of a display screen of the compatible model search tool. As shown in FIG. 7, when the ID assigned to the device to be verified is entered in the input section A and searched, the configuration information linked to the ID and the learned model information matching the configuration information are displayed. be.
  • Fig. 8 is another example of the display screen of the matching model search tool. As shown in FIG. 8, when the ID given to the verification target is entered in the input section A for searching, information of trained model candidates that match the configuration of the verification target is displayed. The information of the learned model candidates in the example of FIG. 8 is output in descending order of compatibility between the model name and the verification target (that is, the numerical value of the degree of compatibility is high).
  • the result reception unit 124 is means for receiving the result of matching the trained model to the verification target.
  • the result accepting unit 124 accepts matching results through an operation from the input device 509 .
  • the result accepting unit 124 accepts the result based on criteria such as higher, equal, or lower than the value of the degree of fitness output by the matching model search tool, and stores the result in the storage device 505 .
  • the model search unit 122 refers to the stored results when searching for a matching model thereafter.
  • model analysis device 120 configured as above will be described with reference to the flowchart of FIG.
  • FIG. 9 is a flow chart showing an overview of the operation of the model analysis device 120 in the second embodiment. Note that the processing according to this flowchart may be executed based on program control by the processor described above.
  • the target information acquisition unit 111 first acquires the configuration information of the device to be verified (step S201).
  • the model search unit 122 searches for a suitable learned model based on the configuration of the device to be verified (step S202).
  • the output unit 203 outputs information on the learned model retrieved by the model retrieval unit 122 (step S203).
  • the result receiving unit 124 receives the input of the result of actually adapting the learned model (step S204). The result received by the result receiving unit 124 is used when the model searching unit 122 searches for a trained model when the flow is repeated again. With this, the model analysis device 100 ends the model analysis operation.
  • the model search unit 122 performs the to search for a matching trained model. As a result, it is possible to obtain information on a suitable learned model by simply inputting device configuration information.
  • a matching trained model is searched based on device configuration information including hardware configuration information to be verified, firmware configuration information, and protocol processing software configuration information.
  • the model search unit 122 for example, based on device configuration information including hardware configuration information to be verified, software configuration information including firmware configuration information and protocol processing software configuration information, network configuration information, and verification content. You can also search for a matching model by
  • target information acquisition means for acquiring device configuration information including hardware configuration information for a verification target and software configuration information including firmware configuration information and protocol processing software configuration information
  • model information acquisition means for acquiring configuration information of the device in an environment in which the learned model used for verification of the verification target is generated; suitability evaluation means for evaluating the suitability of the verification target for the learned model based on the acquired device configuration information of the verification target and the learned model; and output means for outputting the evaluated result.
  • the configuration information of the device in the environment in which the verification target and the trained model are generated includes configuration information of a plurality of devices,
  • the target information acquiring means and the model information acquiring means further acquire, as configuration information, network configuration information including connection information between the plurality of devices,
  • the model analysis device according to appendix 1.
  • Appendix 3 The model analysis device according to appendix 1 or appendix 2, wherein the hardware configuration information for the verification object and the hardware configuration information for which the trained model is generated include chip configuration information for controlling the operation of the device.
  • Appendix 4 further comprising fitness calculation means for calculating a fitness indicating the degree of fitness based on the obtained configuration information of the verification target and the learned model; 4.
  • the model analysis device according to any one of appendices 1 to 3, wherein the suitability evaluation means evaluates the suitability based on the calculated degree of suitability.
  • Appendix 5 further comprising configuration information storage means for storing the identifier information to be verified and the configuration information to be verified in association with each other; 5.
  • the model analysis apparatus according to any one of appendices 1 to 4, wherein the target information acquisition means acquires the configuration information stored in the configuration information storage means based on the identifier information of the verification target.
  • target information acquisition means for acquiring device configuration information including hardware configuration information for a verification target and software configuration information including firmware configuration information and protocol processing software configuration information; model search means for searching for a trained model that conforms to verification of the verification target based on the acquired configuration information of the device; and output means for outputting the retrieved learned model.
  • the configuration information of the device in the verification target includes configuration information of a plurality of devices;
  • the target information acquiring means further acquires network configuration information including connection information between the plurality of devices as configuration information.
  • the model analysis device according to appendix 6.
  • (Appendix 12) Acquiring device configuration information including hardware configuration information in a verification target and software configuration information including firmware configuration information and protocol processing software configuration information, Acquiring configuration information of the device in an environment in which the trained model that matches the verification target is generated; evaluating the suitability of the verification target for the learned model based on the obtained device configuration information of the verification target and the learned model; A model analysis method for outputting the evaluated result.
  • (Appendix 13) Acquiring device configuration information including hardware configuration information in a verification target and software configuration information including firmware configuration information and protocol processing software configuration information, searching for a trained model that conforms to verification of the verification target based on the acquired configuration information of the device; A model analysis method for outputting the retrieved learned model.
  • (Appendix 14) Acquiring device configuration information including hardware configuration information in a verification target and software configuration information including firmware configuration information and protocol processing software configuration information, Acquiring configuration information of the device in an environment in which the trained model that matches the verification target is generated; evaluating the suitability of the verification target for the learned model based on the obtained device configuration information of the verification target and the learned model; A recording medium storing a program that causes a computer to output the evaluated result.
  • (Appendix 15) Acquiring device configuration information including hardware configuration information in a verification target and software configuration information including firmware configuration information and protocol processing software configuration information, searching for a trained model that conforms to verification of the verification target based on the acquired configuration information of the device;
  • a recording medium storing a program that causes a computer to output the retrieved learned model.
  • model analysis device 101 100, 110, 120 model analysis device 101, 111, 121 target information acquisition unit 102, 112 model information acquisition unit 113 suitability calculation unit 103, 114 suitability evaluation unit 104, 115, 123 output unit 122 model search unit 124 result reception Department

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Hardware Design (AREA)
  • Quality & Reliability (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Stored Programmes (AREA)
PCT/JP2021/039981 2021-10-29 2021-10-29 モデル分析装置、モデル分析方法、及び記録媒体 Ceased WO2023073910A1 (ja)

Priority Applications (3)

Application Number Priority Date Filing Date Title
PCT/JP2021/039981 WO2023073910A1 (ja) 2021-10-29 2021-10-29 モデル分析装置、モデル分析方法、及び記録媒体
US18/694,032 US20240385943A1 (en) 2021-10-29 2021-10-29 Model analysis device, model analysis method, and recording medium
JP2023556017A JP7666628B2 (ja) 2021-10-29 2021-10-29 モデル分析装置、モデル分析方法、及びプログラム

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/JP2021/039981 WO2023073910A1 (ja) 2021-10-29 2021-10-29 モデル分析装置、モデル分析方法、及び記録媒体

Publications (1)

Publication Number Publication Date
WO2023073910A1 true WO2023073910A1 (ja) 2023-05-04

Family

ID=86157609

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2021/039981 Ceased WO2023073910A1 (ja) 2021-10-29 2021-10-29 モデル分析装置、モデル分析方法、及び記録媒体

Country Status (3)

Country Link
US (1) US20240385943A1 (https=)
JP (1) JP7666628B2 (https=)
WO (1) WO2023073910A1 (https=)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2017516181A (ja) * 2014-03-13 2017-06-15 クアルコム，インコーポレイテッド 周辺デバイスを安全にするための挙動分析
JP2020067911A (ja) * 2018-10-25 2020-04-30 トヨタ自動車株式会社 制御支援装置、車両、および制御支援システム
JP2020161167A (ja) * 2017-03-21 2020-10-01 株式会社Ｐｒｅｆｅｒｒｅｄ Ｎｅｔｗｏｒｋｓ サーバ装置、学習済モデル提供プログラム、学習済モデル提供方法及び学習済モデル提供システム

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2017516181A (ja) * 2014-03-13 2017-06-15 クアルコム，インコーポレイテッド 周辺デバイスを安全にするための挙動分析
JP2020161167A (ja) * 2017-03-21 2020-10-01 株式会社Ｐｒｅｆｅｒｒｅｄ Ｎｅｔｗｏｒｋｓ サーバ装置、学習済モデル提供プログラム、学習済モデル提供方法及び学習済モデル提供システム
JP2020067911A (ja) * 2018-10-25 2020-04-30 トヨタ自動車株式会社 制御支援装置、車両、および制御支援システム

Also Published As

Publication number Publication date
JPWO2023073910A1 (https=) 2023-05-04
JP7666628B2 (ja) 2025-04-22
US20240385943A1 (en) 2024-11-21

Similar Documents

Publication Publication Date Title
US11171845B2 (en) QoS-optimized selection of a cloud microservices provider
US9990501B2 (en) Diagnosing and tracking product vulnerabilities for telecommunication devices via a database
US9619655B2 (en) Cloud-based security profiling, threat analysis and intelligence
WO2020139635A1 (en) Automated training and selection of models for document analysis
US20210365511A1 (en) Generation and delivery of content curated for a client
GB2503486A (en) Managing changes to files
US12602597B2 (en) Automatic discovery of machine learning model features
KR101525984B1 (ko) 파일 이름을 규칙에 따라 설정하는 문서 관리 장치 및 방법
US20220103586A1 (en) Tailored network risk analysis using deep learning modeling
US20170322787A1 (en) Providing enterprise product bundles as a cloud-based service
US20120023081A1 (en) Customizing search home pages using interest indicators
US11144930B2 (en) System and method for managing service requests
CN108885444A (zh) 信息管理装置、信息管理方法及信息管理系统
US20230101955A1 (en) Reuse of machine learning models
KR20110139896A (ko) 금융 상품 추천 방법
JP5962736B2 (ja) 情報処理システム、分類方法、及びそのためのプログラム
US20080005070A1 (en) Non-Repetitive Web Searching
WO2023073910A1 (ja) モデル分析装置、モデル分析方法、及び記録媒体
CN114089997A (zh) 页面栈的处理方法和装置、存储介质及电子装置
US20120317526A1 (en) Verification method, computer-readable recording medium, and design verification apparatus
US12363158B1 (en) Generating action recommendations based on attributes associated with incidents used for incident response
CN104111977B (zh) 信息匹配方法、装置及终端
JP2010282503A (ja) 関連情報検索表示システム及び関連情報検索表示方法，プログラム
KR102620571B1 (ko) 콘텐츠 추천 방법 및 시스템
US8964730B2 (en) IP telephone provisioning system and methods

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21962452

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 18694032

Country of ref document: US

ENP Entry into the national phase

Ref document number: 2023556017

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21962452

Country of ref document: EP

Kind code of ref document: A1