Information processing method and information processing system
The method protects machine learning parameters by using TEEs for parameter concealment and input anonymization within a neural network, addressing the leakage issues in conventional confidential information processing technologies and ensuring secure handling of sensitive data.
Patent Information
- Application Number
- PCT/JP2024/017207
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-15
- Filing Date
- 2024-05-09
- Publication Date
- 2024-11-21
AI Technical Summary
Conventional technologies for confidential information processing, such as those using Trusted Execution Environments (TEEs), struggle to protect machine learning parameters during deep learning processing, particularly when handling unknown input information containing personal privacy data or trade secrets, as they cannot prevent leakage of deep learning model parameters.
An information processing method employing a neural network with multiple layers, utilizing a Trusted Execution Environment (TEE) to execute parameter concealment and input masking processes, generating concealed parameters and anonymized input data, and performing processing in both isolated and normal areas to prevent information leakage.
Effectively protects machine learning parameters and prevents information leakage during processing, ensuring confidentiality even when using high-speed processing areas like GPUs, thereby enhancing the security of confidential information handling.
Smart Images

Figure JP2024017207_21112024_PF_FP_ABST
Abstract
Description
Information processing method and information processing system
[0001] The present invention relates to a secret information processing method and secret information processing.
[0002] In recent years, there has been progress in schemes for utilizing personal data through collaboration between the public and private sectors. For example, new value is being created by linking diverse information, such as through collaboration between public and private services. Statistical analysis such as regression analysis and machine learning such as CNN (Convolutional Neural Network) are considered effective in creating new value by linking diverse information. Platforms are also provided that enable various companies to easily implement statistical analysis and machine learning using inexpensive public clouds.
[0003] On the other hand, regulations aimed at protecting privacy are becoming stricter. Therefore, there is a need for technology that can utilize confidential information held by data holders such as hospitals and banks while keeping it confidential.
[0004] Against this backdrop, the creation of new value through the integration of diverse information is being promoted, such as public services that share information between the public and private sectors, such as emergency services and security companies, and collaboration between private services such as pharmaceuticals, insurance, transportation, and information banks, which create synergistic effects with public services. Statistical analysis such as regression analysis and analytical processing such as machine learning are considered effective for analyzing diverse data and creating new value. Furthermore, to realize these services, confidential information processing technology is required that enables public and private service providers to analyze and utilize confidential information, such as personal information, held by data holders such as hospitals and banks, while protecting the privacy of individuals.
[0005] Conventional technologies enable specific processing of encrypted data, thereby enabling the use of confidential information in a confidential state. However, the processes that can be performed with conventional technologies are limited to basic processes such as search and order comparison, and the degree of freedom of processing is limited. Therefore, there is a need for technology that can handle advanced statistical processing and image processing using machine learning such as CNN, which require a high degree of processing freedom.
[0006] In recent years, central processing units (CPUs) have been equipped with a trusted execution environment (TEE) function. The TEE function is considered to be effective for processing confidential information. The TEE function provides a trusted area where information cannot be read even if the administrator privileges of the OS running on the computer are stolen. By decrypting and processing encrypted data within the trusted area, secure and advanced processing becomes possible.
[0007] A technology that utilizes the trust region of the TEE function is described in Patent Document 1. Patent Document 1 discloses a privacy-enhanced deep learning system that reduces information leakage of confidential input data in an inference pipeline.
[0008] The technology described in Patent Document 1 utilizes the hierarchical structure of a neural network to divide each deep learning model into a FrontNet that processes in the trusted region and a BackNet that processes in the normal region.
[0009] In the technology described in Patent Document 1, a terminal used by an end user transmits encrypted input and encrypted FrontNet to a system. The technology described in Patent Document 1 utilizes the TEE function on the cloud infrastructure to perform deep learning processing in the FrontNet enclave (an isolated execution process in the TEE function), and cryptographically protects the confidentiality and integrity of user input.
[0010] On the other hand, in the technology described in Patent Document 1, there is a lack of secure enclaves for BackNet inference calculations, and if the cloud machine is equipped with a deep learning acceleration chip such as a GPU (Graphical Processing Unit), the benefits of improved performance can be obtained.
[0011] The technology described in Patent Literature 1 utilizes the hierarchical structure of a neural network and partitions each deep learning model into a FrontNet and a BackNet to determine a model-specific optimal split point that balances privacy protection and performance requirements. In the technology described in Patent Literature 1, a deep learning inference system measures the similarity for each set of intermediate data in each layer and determines the split point to be used for dividing into two partitions using a selected subset of intermediate data that is most similar to the input information.
[0012] Furthermore, a technology that utilizes the trust region of the TEE function is described in Non-Patent Document 1. Non-Patent Document 1 discloses a method for concealing confidential input data that reduces information leakage of confidential input data input to a GPU in an inference pipeline that utilizes a GPU that speeds up linear arithmetic processing in deep learning.
[0013] US Patent Application Publication No. 2020 / 0082259
[0014] F. Tramer and D. Boneh, “Slalom: Fast, verifiable and private execution of neural networks in trusted hardware,” in International Conference on Learning Representations (ICLR), 2019.
[0015] However, the system described in Patent Literature 1 may not be able to prevent the leakage of confidential information during deep learning processing of unknown input information if the unknown input information includes confidential information such as personal privacy information or trade secrets that have not been learned in the learning stage. Specifically, BackNet processes the machine learning parameters that make up the deep learning model in the normal domain, so it cannot prevent the leakage of the deep learning model if the machine learning parameters are confidential information.
[0016] Furthermore, the concealment method described in Non-Patent Document 1 processes machine learning parameters used in linear arithmetic processing, which is part of a deep learning model, without concealment on a normal GPU, and therefore cannot prevent leakage of confidential information when the machine learning parameters are confidential information.
[0017] Therefore, one aspect of the present invention aims to protect machine learning parameters, which are confidential information, during processing such as machine learning using multiple processing layers.
[0018] In order to solve the above-mentioned problems, one aspect of the present invention employs the following configuration: That is, an information processing method performed by an information processing system that executes information processing using a neural network consisting of multiple i-th layers (i is a natural number between 1 and N), the information processing system having a processor and a memory, and the memory being a Trusted Execution Environment (TEE) isolated from a normal area. the processor executes, for each layer i, a parameter concealment process in the isolation area, which multiplies the layer i parameter by the layer i parameter protection random number to generate a layer i concealment parameter by concealing the layer i parameter, outputs the layer i concealment parameter to the normal area, and generates an input concealment effect removal parameter by multiplying the layer i parameter by the layer i input data protection random number to remove the concealment applied to the layer i input data, and stores the input concealment effect removal parameter in the isolation area. the layer-i input data protection random number is added to the layer-i input data to generate concealed input data; the layer-i input data is output to the normal area as the layer-i input data; the layer-i concealed input data is obtained from the normal area as a result of the layer-i processing performed on the layer-i concealed input data based on the layer-i concealment parameter in the normal area; the layer-i parameter is used to obtain the execution result ...
[0019] According to one aspect of the present invention, machine learning parameters, which are confidential information, can be protected during processing such as machine learning using multiple processing layers.
[0020] Problems, configurations, and effects other than those described above will become apparent from the following description of the embodiments.
[0021] 1 is a block diagram showing an example of a system configuration of a confidential information processing system. FIG. 2 is a block diagram showing an example of a hardware configuration of a confidential information processing system. FIG. 3 is an explanatory diagram showing an example of a functional configuration and an example of a data processing flow for confidential inference processing. FIG. 4 is a sequence diagram showing an example of a data processing flow of confidential inference preparation processing. FIG. 5 is a sequence diagram showing an example of a data processing flow of confidential inference execution processing. FIG. 6 is a flowchart showing an example of a parameter concealment processing. FIG. 7 is a flowchart showing an example of an input concealment influence removal parameter generation processing. FIG. 8 is a flowchart showing an example of a concealment input data generation processing. FIG. 9 is a flowchart showing an example of a parameter concealment influence removal processing. FIG. 10 is a dialog diagram showing an example of a machine learning parameter concealment setting information input screen. FIG. 11 is a table showing an example of a each layer processing concealment setting information table.
[0022] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. In this embodiment, the same components are generally designated by the same reference numerals, and repeated explanations will be omitted. It should be noted that this embodiment is merely an example for realizing the present invention, and does not limit the technical scope of the present invention.
[0023] (1) Configuration Example of Confidential Information Processing System 1 Fig. 1 is a block diagram showing an example system configuration of the confidential information processing system 1. The confidential information processing system 1 is entrusted with encrypted machine learning parameters and the like from a parameter holder that holds confidential information that is a trade secret, such as machine learning parameters. The confidential information processing system 1 also receives encrypted data from a data holder that holds confidential information, such as personal information. The confidential information processing system 1 processes the entrusted machine learning parameters and the like and data in accordance with a processing request from a processing result user while keeping the confidential information confidential, and provides the processing result to the processing result user.
[0024] The confidential information processing system 1 includes, for example, a parameter holder terminal 50 used by a parameter holder to encrypt and deposit machine learning parameters, etc. The confidential information processing system 1 also includes a data holder terminal 10 used by a data holder to encrypt and deposit data. The confidential information processing system 1 also includes a processing result user terminal 20 through which a processing result user generates and transmits a processing request and references the processing results. The confidential information processing system 1 also includes a confidential information processing server 30 that executes processing of a processing request received from a processing result user on data deposited from the data holder while keeping it confidential based on the machine learning parameters, etc. deposited from the parameter holder, and transmits the processing results. The parameter holder terminal 50, the data holder terminal 10, the processing result user terminal 20, and the confidential information processing server 30 are connected to each other via a network 40 such as the Internet.
[0025] The confidential information processing system 1 may include a plurality of parameter holder terminals 50, a plurality of data holder terminals 10, a plurality of processing result user terminals 20, or a plurality of confidential information processing servers. The parameter holder terminal 50 and the data holder terminal 10 may be the same terminal, or the data holder terminal 10 and the processing result user terminal 20 may be the same terminal.
[0026] 2 is a block diagram showing an example of the hardware configuration of the secret information processing system 1. The data holder terminal 10 is, for example, a computer such as a personal computer, a smartphone, or a server device, or a virtual computer. The data holder terminal 10 is configured by a computer including a CPU (Central Processing Unit) 11, a memory 12, an auxiliary storage device 13, a network interface 14, a display device 15, and an input device 16, which are connected to each other by, for example, an internal communication line.
[0027] The CPU 11 includes a processor and executes programs stored in the memory 12. The memory 12 includes a non-volatile storage element, Read Only Memory (ROM), and a volatile storage element, Random Access Memory (RAM). The ROM stores unchanging programs (e.g., a Basic Input / Output System (BIOS)). The RAM is a high-speed, volatile storage element such as a Dynamic Random Access Memory (DRAM), and temporarily stores programs executed by the CPU 11 and data used when the programs are executed.
[0028] The CPU 11 is a TEE-enabled CPU having a TEE (Trusted Execution Environment) function, and the memory 12 has a TEE trusted area 17, which is a hardware trusted area secured on the memory by the TEE-enabled CPU 21 and is isolated from other areas on the memory. In the first embodiment, the CPU 11 does not need to have the TEE function, and the memory 12 does not need to have the TEE trusted area 17. The memory capacity that can be handled in the TEE trusted area 17 may be limited to be smaller than the memory capacity of the normal area (for example, about 128 MB). In this case, when the CPU 11 executes processing on the TEE trusted area 17, the processing speed decreases compared to when the CPU 11 executes processing on the normal area.
[0029] The auxiliary storage device 13 is a large-capacity, non-volatile storage device such as a magnetic storage device (HDD (Hard Disk Drive)) or a flash memory (SSD (Solid State Drive)), and stores programs to be executed by the CPU 11 and data to be used when the programs are executed. That is, the programs are read from the auxiliary storage device 13, loaded into the memory 12, and executed by the CPU 11.
[0030] The display device 15 is a device such as a display or printer that outputs the results of program execution in a format that can be viewed by an operator. The input device 16 is a device such as a keyboard or mouse that receives input from an operator. The network interface 14 controls communication with other devices in accordance with a predetermined protocol. The network interface 14 may also include a serial interface such as a USB.
[0031] The program executed by the CPU 11 may be stored in advance in the non-volatile auxiliary storage device 13, which is a computer-readable non-transitory storage medium, or in removable media (CD-ROM, flash memory, etc.), which are non-transitory storage media. Alternatively, the program may be provided to the data holder terminal 10 from a non-transitory storage device of another device via a network and stored in the auxiliary storage device 13. For this reason, it is preferable that the data holder terminal 10 has an interface for reading data from removable media. This also applies to the processing result user terminal 20 and the confidential information processing server 30.
[0032] In this embodiment, some or all of the functions of the functional units executed by the CPU and GPU may be realized by hardware such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field-Programmable Gate Array).
[0033] In this embodiment, the information used by the confidential information processing system 1 can be expressed in any data structure regardless of the data structure, and the information can be stored in a data structure appropriately selected from, for example, a list, a table, a database, or a queue.
[0034] The parameter holder terminal 50 is, for example, a computer such as a personal computer, a smartphone, or a server device, or a virtual computer. The parameter holder terminal 50 is configured by a computer including, for example, a CPU 51, a memory 52, an auxiliary storage device 53, a network interface 54, a display device 55, and an input device 56, which are connected to each other by an internal communication line.
[0035] The CPU 51, memory 52, auxiliary storage device 53, network interface 54, display device 55, and input device 56 are the same hardware as the CPU 11, memory 12, auxiliary storage device 13, network interface 14, display device 15, and input device 16. However, the CPU 51 does not need to have a TEE function, and the memory 52 does not need to have a trusted region.
[0036] The processing result user terminal 20 is, for example, a computer such as a personal computer, a smartphone, or a server device, or a virtual computer. The processing result user terminal 20 is configured by, for example, a computer including a CPU 21, a memory 22, an auxiliary storage device 23, a network interface 24, a display device 25, and an input device 26, which are connected to each other via an internal communication line.
[0037] The CPU 21, memory 22, auxiliary storage device 23, network interface 24, display device 25, and input device 26 are the same hardware as the CPU 11, memory 12, auxiliary storage device 13, network interface 14, display device 15, and input device 16. However, the CPU 21 does not need to have a TEE function, and the memory 22 does not need to have a trusted region.
[0038] The confidential information processing server 30 is, for example, a computer such as a personal computer, a smartphone, or a server device, or a virtual computer. The confidential information processing server 30 is configured by a computer including a CPU 31, a memory 32, an auxiliary storage device 33, a network interface 34, a display device 35, an input device 36, and a GPU (Graphics Processing Unit) 39, which are connected to each other by an internal communication line such as a bus.
[0039] The CPU 31, memory 32, auxiliary storage device 33, network interface 34, display device 35, and input device 36 are the same hardware as the CPU 11, memory 12, auxiliary storage device 13, network interface 14, display device 15, and input device 16. However, the CPU 31 of the secret information processing server 30 is a TEE-enabled CPU with a TEE function. The memory 32 of the secret information processing server 30 is a hardware-based trusted area secured on the memory by the TEE-enabled CPU 31, and has a TEE trusted area 38, which is an execution area isolated from other areas on the memory. The TEE trusted area 38 is also referred to as an isolated area isolated from the normal area. The GPU 39 includes a processor and executes programs stored in the memory 32 at high speed using an area (normal area) of the memory 32 outside the TEE trusted area 38. Note that the secret information processing server 30 does not necessarily have to have the GPU 39.
[0040] In this embodiment, the parameter holder terminal 50, the data holder terminal 10, and the confidential information processing server 30 have a TEE function, but a method other than the TEE function may also be adopted, which allows calculations to be performed in a secure execution area isolated from the normal area on the memory.
[0041] In addition, some or all of the processing performed by CPU 11, CPU 21, CPU 31, CPU 51, and GPU 39 may be performed by hardware such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field-Programmable Gate Array).
[0042] The parameter holder terminal 50, the data holder terminal 10, the processing result user terminal 20, and the confidential information processing server 30 are all computer systems configured on one physical computer or on multiple logically or physically configured computers. However, the parameter holder terminal 50, the data holder terminal 10, the processing result user terminal 20, and the confidential information processing server 30 may run in separate threads on the same computer, or may run on virtual computers built on multiple physical computer resources.
[0043] The network 40 is a communication network such as the Internet via wired or wireless communication, an intranet such as an in-house network, or a satellite line.
[0044] (2) Example of Functional Configuration of Confidential Information Processing System 1 and Example of Data Processing of Confidential Inference Processing FIG. 3 is an explanatory diagram showing an example of a functional configuration and an example of a data processing flow for confidential inference processing.
[0045] First, an overall description will be given of the confidential inference processing executed by the confidential information processing system 1 according to this embodiment. In the confidential information processing system 1, the parameter holder terminal 50 first encrypts machine learning parameter information including confidential information such as each layer processing concealment setting information, and transmits the encrypt information to the confidential information processing server 30. The each layer processing concealment setting information is information that specifies an area in which machine learning parameters of a neural network such as a CNN (Convolutional Neural Network) and processing of each layer are executed while being kept confidential.
[0046] Next, the confidential information processing server 30 decrypts the machine learning parameter information in the TEE trust region 38, and stores the parameter concealment setting information in the TEE trust region 38. Then, the parameter concealment processing and the input concealment influence removal parameter generation processing are executed in the TEE trust region 38, the concealed machine learning parameter information is output to and stored in the normal region, and the input concealment removal parameter is stored in the TEE trust region 38.
[0047] Next, in the secret information processing system 1 , the data holder terminal 10 encrypts input data including confidential information such as personal information and transmits the encrypted data to the secret information processing server 30 .
[0048] Next, the confidential information processing server 30 decrypts the input data in the TEE trusted region 38, and executes the processing of each layer of the deep learning inference processing including multiple processing layers in the TEE trusted region 38 or the normal region depending on the result of the anonymization setting confirmation processing based on the parameter anonymization setting information, and encrypts the inference result and transmits it to the processing result user terminal 20. Then, the processing result user terminal 20 decrypts the received encrypted inference result and acquires the inference result.
[0049] An example of the functional configuration and an example of the data processing flow for each of the above-mentioned processes will be described in detail below with reference to FIG.
[0050] The confidential information processing server 30 has, for example, a normal area 310 on the memory 32 for performing high-speed processing using a normal CPU and memory, and an acceleration chip such as a GPU. The confidential information processing server 30 also has a TEE trusted area 38 on the memory 32, which is a highly securely isolated processing execution area provided by the TEE function of the CPU or the like.
[0051] The normal region 310 includes, for example, an i-th layer processing unit 312. The i-th layer processing unit 312 executes processing of each layer in machine learning with a multi-layer structure such as a CNN (hereinafter, the description will be given using a CNN as an example). Note that a CNN is merely an example, and this embodiment can also be applied to other neural networks such as a recurrent neural network (RNN).
[0052] The i-th layer processing unit 312 performs layer processing of the anonymized input data 516 based on the anonymized machine learning parameters 515 to generate anonymized intermediate data 517, and inputs it to the parameter anonymization influence removal processing unit 533 of the TEE trust region 38.
[0053] The TEE trust region 38 includes, for example, an i-th layer processing unit 326, a decryption processing unit 522, a decryption processing unit 322, a parameter concealment processing unit 534, an input concealment influence removal parameter generation processing unit 535, and an i-th layer concealment setting confirmation processing unit 536. The TEE trust region 38 also includes a concealed input data generation processing unit 525, a parameter concealment influence removal processing unit 533, an input concealment influence removal processing unit 538, a random number generation unit 539, and an encryption processing unit 329.
[0054] The i-th layer processing unit 326 performs processing of the CNN layer. The random number generation unit 539 generates random numbers used in various anonymization processes. The decryption processing unit 522 decrypts the encrypted machine learning parameter information 521 (including machine learning parameters and each layer processing anonymization setting information) using the encryption key 532 stored in the TEE trust region 38. The decryption processing unit 522 stores the decrypted machine learning parameter information 521 in the TEE trust region 38 as machine learning parameters 523 and each layer processing anonymization setting information 524. The decryption processing unit 322 decrypts the encrypted data 103 using the encryption key 331 stored in the TEE trust region 38, and inputs the decrypted data 103 to the i-th layer anonymization setting confirmation processing unit 536.
[0055] Parameter anonymization processing unit 534 receives as input machine learning parameters 523 of each layer included in machine learning parameters 523 and parameter protection random numbers 528 that protect different parameters for each layer generated by random number generation unit 539, and generates anonymized machine learning parameters 515. Parameter anonymization processing unit 534 stores generated anonymized machine learning parameters 515 in general area 310.
[0056] The input concealment influence removal parameter generation processing unit 535 receives as input the machine learning parameters 523 of each layer and the input data protection random numbers 526 that protect the input data of each layer, which differ for each layer and are generated by the random number generation unit 539, and generates an input concealment influence removal parameter 529. The input concealment influence removal parameter generation processing unit 535 stores the generated input concealment influence removal parameter 529 in the TEE trust region 38.
[0057] The i-th layer concealment setting confirmation processing unit 536 receives as input the input data of the layer (input data 323 or intermediate data 530) and the layer number i of the processing layer. The i-th layer concealment setting confirmation processing unit 536 references information on the region in which the i-th layer processing of the layer number is executed in each layer processing concealment setting information 524, and if the reference result indicates a "trusted region," it inputs the input data to the i-th layer processing unit 326. On the other hand, if the reference result indicates a "normal region," the i-th layer concealment setting confirmation processing unit 536 inputs the input data to the concealed input data generation processing unit 525.
[0058] The i-th layer processing unit 326 executes processing of a layer in a multi-layer machine learning structure such as a CNN within the TEE trust region 38, and outputs the execution result as intermediate data 327 if it is not the final layer, and outputs the execution result as an inference result 328 if it is the final layer.
[0059] The concealed input data generation processor 525 receives the input data of the layer in question as input, reads the random number of the layer in the input data protection random number 526, and reads the temporary input data protection random number 537 generated by the random number generator 539 for each concealed input data generation process. The concealed input data generation processor 525 executes a process of concealing the input data based on these random numbers, and outputs the concealed input data 516. Note that the concealed input data generation processor 525 may perform concealment without using the temporary input data protection random number 537.
[0060] The parameter concealment influence removal processing unit 533 performs a parameter concealment influence removal processing on the input concealment intermediate data 517, based on the parameter protection random number 528 of each layer, to remove the influence of parameter concealment from the concealment intermediate data 517. The parameter concealment influence removal processing unit 533 inputs the execution result of the parameter concealment influence removal processing to the input concealment influence removal processing unit 538.
[0061] The input concealment influence elimination processor 538 executes input concealment influence elimination processing on input data, removing the influence of concealment on the input data to the layer, based on the input concealment influence elimination parameter 529 of the layer and the temporary input data protection random number 537. The input concealment influence elimination processor 538 outputs an inference result 531 if the layer is other than the final layer, and inputs the inference result 531 to the encryption processor 329 if the layer is the final layer. Note that if concealment is performed without using the temporary input data protection random number 537 in the concealment input data generation processing for the layer, the temporary input data protection random number 537 is not used in the input concealment influence elimination processing. The temporary input data protection random number 537 may also be generated in advance at the stage of concealment inference preparation processing.
[0062] The encryption processing unit 229 performs encryption on the input inference result 328 or inference result 531 using the encryption key 332 stored in the TEE trusted area 38, and outputs the encrypted inference result 330 via the normal area 310.
[0063] The parameter holder terminal 50 also includes an encryption processing unit 502 that encrypts machine learning parameter information 501, etc. using an encryption key 504 stored in the memory 52, the auxiliary storage device 53, and the TEE trusted area 57. The data holder terminal 10 includes an encryption processing unit 102 that encrypts data using an encryption key 104 stored in the memory 12, the auxiliary storage device 13, or the TEE trusted area 17. The processing result user terminal 20 includes a decryption processing unit 202 that decrypts encrypted inference results, etc. using an encryption key 204 stored in the memory 22 or the auxiliary storage device 23. Note that any two or all of the parameter holder terminal 50, the data holder terminal 10, and the processing result user terminal may be the same terminal, in which case any two or all of the encryption processing unit 502, the encryption processing unit 102, and the decryption processing unit 202 are included in the same terminal.
[0064] Moreover, the GPU 39 operates in accordance with the i-th layer processing program loaded into the normal area of the memory 32, thereby performing processing as the i-th layer processing unit 312. Moreover, the TEE-enabled CPU 51 operates in accordance with the encryption processing program loaded into the TEE trusted area 57 of the memory 52, thereby performing processing as the encryption processing unit 502. Moreover, the TEE-enabled CPU 11 operates in accordance with the encryption processing program loaded into the TEE trusted area 17 of the memory 12, thereby performing processing as the encryption processing unit 102. Moreover, the CPU 21 operates in accordance with the decryption processing program loaded into the memory 22, thereby performing processing as the decryption processing unit 202.
[0065] The data processing flow of the confidential inference preparation processing according to the first embodiment will be described below with reference to Figures 3 and 4. Figure 4 is a sequence diagram showing an example of the data processing flow of the confidential inference preparation processing. First, the CPU 51 of the parameter holder terminal 50 displays the machine learning parameter anonymization setting information input screen shown in Figure 11 on the display device 55 in accordance with the machine learning parameter anonymization setting information input screen display program loaded into the memory 52. The CPU 51 then accepts input from the user of the name of the target machine learning parameter, the number of layers of the target machine learning parameter, whether the target machine learning parameter needs to be anonymized, and each layer processing anonymization setting information for the target machine learning parameter (step S401).
[0066] Next, the encryption processing unit 502 of the parameter holder terminal 50 reads and encrypts machine learning parameters with names based on the user's input shown in FIG. 11 from the auxiliary storage device 53 (step S402). The encryption processing unit 502 then reads and encrypts each layer processing concealment setting information based on the user's input from the memory 52 (step S403). The encryption processing unit 502 then registers the machine learning parameters and each layer processing concealment setting information by transmitting the encrypted machine learning parameter information to the confidential information processing server 30 (step S404). Upon receiving the encrypted data 503, the confidential information processing server 30 inputs it into the TEE trust region 38. The decryption processing unit 522 of the TEE trust region 38 decrypts the encrypted data 503 to generate machine learning parameters 523 and each layer processing concealment setting information 524 having the data structure shown in FIG. 12 (step S405).
[0067] Next, the parameter anonymization processing unit 534 executes the parameter anonymization processing shown in FIG. 6, outputs the execution result to the general area 310 (step S406), and stores it in the general area 310 as anonymized machine learning parameters 515 (step S407).
[0068] Next, the input concealment effect removal parameter generation processing unit 535 executes the input concealment effect removal parameter generation processing shown in FIG. 7 (step S408), stores the execution result as the input concealment effect removal parameter 529 in the TEE trust region 38, and transmits a registration completion response to the parameter holder terminal 50 (step S409).
[0069] The data processing flow of the confidential inference execution process according to the first embodiment will be described below with reference to Figures 3 and 5. Figure 5 is a sequence diagram showing an example of the data processing flow of the confidential inference execution process. First, the encryption processing unit 102 of the data holder terminal 10 encrypts data 101 including confidential information to generate encrypted data 103 (step S501), and registers the data by transmitting the encrypted data 103 to the confidential information processing server 30 (step S502). Upon receiving the encrypted data 103, the confidential information processing server 30 inputs it into the TEE trust region 38. The decryption processing unit 322 of the TEE trust region 38 decrypts the encrypted data 103 to generate input data 323 (step S503), and initializes the value of a variable i indicating the layer number of the process to 1 (step S504).
[0070] The i-th layer concealment setting confirmation processing unit 536 reads information on the region in which the i-th layer processing is to be performed from the layer processing concealment setting information 524 (step S505). The i-th layer concealment setting confirmation processing unit 536 then proceeds to step S506. If the read result indicates a "trusted region" (step S506: No), the i-th layer concealment setting confirmation processing unit 536 inputs the input data to the i-th layer processing unit 326 and reads out the i-th layer machine learning parameter information (step S507). The i-th layer concealment setting confirmation processing unit 536 then executes the i-th layer processing within the TEE trusted region 38 (step S508).
[0071] On the other hand, if the read result indicates the "normal area" (Yes in step S506), the i-th layer anonymization setting confirmation processing unit 536 inputs the input data to the anonymized input data generation processing unit 525 and executes the anonymized input data generation processing shown in Fig. 8 (step S509). Then, the i-th layer anonymization setting confirmation processing unit 536 outputs the i-th layer anonymized input data resulting from the execution to the normal area 310 (step S510).
[0072] Furthermore, the layer-i concealment setting confirmation processing unit 536 transitions to step S517 after step S508, and if the layer-i is not the last processing layer (step S517 No), it uses the output data of the layer-i processing as input data for the next layer, increments i, and transitions to step S505. On the other hand, if the layer-i concealment setting confirmation processing unit 536 determines that the layer-i is the last processing layer (step S517 Yes), the encryption processing unit 329 encrypts the output data of the layer-i processing as an inference result and transmits the processing result to the user terminal 20 (step S518).
[0073] Next, the i-th layer processing unit 312 of the normal region 310 reads the i-th layer concealed input data 516 as input (step S511). Next, the i-th layer processing unit 312 executes the i-th layer processing and inputs the execution result as concealed intermediate data 517 to the parameter concealment influence removal unit 533 of the TEE trust region 38 (step S512).
[0074] For example, the i-th layer processing is performed using equation (1). Equation (1) represents an operation of multiplying the i-th layer concealed input data x_i′ matrix by the i-th layer concealed machine learning parameter W_i′ matrix to calculate the execution result y_i′.
[0075]
[0076] Next, the parameter concealment influence removal processing unit 533 executes the parameter concealment influence removal processing shown in FIG. 9 and inputs the parameter concealment influence removal intermediate data of the i-th layer, which is the execution result, to the input concealment influence removal processing unit 538 (step S513).
[0077] Next, the input concealment influence removal processing unit 538 executes the input concealment influence removal processing shown in FIG. 10 (step S514), and proceeds to step S515. If the i-th layer is not the last processing layer (step S515: No), the input concealment influence removal processing unit 538 sets the output data of the i-th layer, which is the execution result, as the input data of the next layer, increments i, and proceeds to step S505. On the other hand, if the i-th layer is the last processing layer (step S515: Yes), the input concealment influence removal processing unit 538 inputs the output data of the i-th layer, which is the execution result, as the inference result to the encryption processing unit 329. Then, the input concealment influence removal processing unit 538 transmits the encrypted inference result to the processing result user terminal 20 (step S518).
[0078] 6 is a flowchart showing an example of the parameter concealment process. The parameter concealment process unit 534 of the TEE trust region 38 first initializes the value of i to 1 (step S601).
[0079] Next, the parameter anonymization processing unit 534 reads information on the region where the i-th layer processing is to be performed from the each layer processing anonymization setting information 524 (step S602), and proceeds to step S603. If the read result indicates a "trusted region" (step S603: No), the parameter anonymization processing unit 534 proceeds to step S607. On the other hand, if the read result indicates a "normal region" (step S603: Yes), the parameter anonymization processing unit 534 proceeds to step S604.
[0080] Next, the parameter concealment processing unit 534 reads the machine learning parameters of the i-th layer from the machine learning parameter 523 (step S604).The parameter concealment processing unit 534 then obtains the random numbers for parameter protection of the i-th layer generated by the random number generation unit 539, and adds and stores them in the parameter protection random numbers 528 (step S605).
[0081] For example, the machine learning parameters of the i-th layer are expressed using equation (2). W_i in equation (2) represents a matrix with p rows and q columns. Furthermore, for example, the i-th layer parameter protection random number for protecting the parameters of the i-th layer is expressed using equation (3). β_i in equation (3) is a scalar random number.
[0082]
[0083]
[0084] Next, the parameter concealment processing unit 534 multiplies the machine learning parameter of the i-th layer by the machine learning parameter of the i-th layer to calculate the concealed machine learning parameter of the i-th layer (step S606).
[0085] For example, the concealed machine learning parameters are calculated using equation (4). W_i′ in equation (4) is an example of an i-th layer concealment parameter, which is an anonymized i-th layer machine learning parameter, and represents a matrix with p rows and q columns.
[0086]
[0087] Next, if the i-th layer is not the last processing layer (step S607: No), the parameter anonymization processing unit 534 increments i (step S608) and proceeds to step S602. On the other hand, if the i-th layer is the last processing layer (step S607: Yes), the parameter anonymization processing unit 534 outputs the calculated set of anonymized machine learning parameters to the general area 310, stores them as anonymized machine learning parameters 515, and terminates the parameter anonymization processing.
[0088] 7 is a flowchart showing an example of input concealment influence elimination parameter generation processing. The input concealment influence elimination parameter generation processing unit 535 of the TEE trust region 38 first initializes the value of i to 1 (step S701).
[0089] Next, the input concealment influence removal parameter generation processor 535 reads information on the region where the i-th layer processing is to be executed from the each layer processing concealment setting information 524 (step S702), and proceeds to step S703. If the read result indicates a "trusted region" (step 703: No), the input concealment influence removal parameter generation processor 535 proceeds to step S707. On the other hand, if the read result indicates a "normal region" (step 703: Yes), the input concealment influence removal parameter generation processor 535 proceeds to step S704.
[0090] Next, the input concealment influence removal parameter generation processing unit 535 acquires the input data protection random number for input data protection of the i-th layer generated by the random number generation unit 539, and adds and stores it to the input data protection random number 526 of the TEE trust region 38 (step S704). Next, the input concealment influence removal parameter generation processing unit 535 reads the machine learning parameters of the i-th layer from the machine learning parameters 523 (step S705).
[0091] For example, the input data protection random number of the i-th layer is expressed using formula (5): r_i in formula (5) is an example of the input data protection random number of the i-th layer, and represents a matrix of m rows and n columns, where each element of the matrix is a different random number.
[0092]
[0093] Furthermore, for example, the input data protection random number of the i-th layer is expressed using equation (6). r_i^(j) in equation (6) is an example of the input data protection random number of the i-th layer, and represents a matrix group with m rows and n columns, and each element of the matrix is a different random number.
[0094]
[0095] Next, the input concealment influence removal parameter generation processing unit 535 multiplies the input data protection random number of the i-th layer by the machine learning parameter of the i-th layer to calculate the input concealment influence removal parameter of the i-th layer (step S706).
[0096] For example, the input concealment effect removal parameter of the i-th layer is calculated using equation (7). u_i in equation (7) is an example of an i-th layer input data concealment removal parameter that removes the influence of concealment from the input data of the i-th layer, and is calculated by multiplying the input data protection random number r_i matrix of the i-th layer and the machine learning parameter W_i matrix of the i-th layer.
[0097]
[0098] Furthermore, for example, an input concealment effect removal parameter for the i-th layer is calculated using equation (8). u_i^(j) in equation (8) is an example of an input concealment effect removal parameter group for the i-th layer, and is calculated by multiplying the input data protection random number r_i^(j) matrix for the i-th layer and the machine learning parameter W_i matrix for the i-th layer.
[0099]
[0100] Next, in step S707, if the i-th layer is not the last processing layer (step S707 No), the input concealment influence removal parameter generation processing unit 535 increments i (step S708) and proceeds to step S702. On the other hand, if the i-th layer is the last processing layer (step S707 Yes), the input concealment influence removal parameter generation processing unit 535 saves the calculated input concealment influence removal parameter group as the input concealment influence removal parameters 529 of the TEE trust region 38 (step S707). When step S707 ends, the input concealment influence removal parameter generation processing unit 535 ends the input concealment influence removal parameter generation processing.
[0101] 8 is a flowchart showing an example of the concealed input data generation process. The concealed input data generation processing unit 525 of the TEE trust region 38 first reads out input data of the i-th layer, which is the layer in question (step S801).
[0102] For example, input data of the i-th layer is expressed using equation (9): x_i in equation (9) is an example of input data of the i-th layer, and represents a matrix with m rows and n columns, and each element of the matrix is a numerical value.
[0103]
[0104] Next, the anonymized input data generation processing unit 525 reads the input data protection random number of the i-th layer from the input data protection random number 526 (step S802), obtains a new random number from the random number generation unit 539, and stores it as a temporary input data protection random number 537 (step S803).
[0105] For example, the temporary input data protection random number of the i-th layer is expressed using equation (10): α_i in equation (10) is an example of the temporary input data protection random number of the i-th layer, and represents a single random number.
[0106]
[0107] Furthermore, for example, the temporary input data protection random numbers of the i-th layer are expressed using equation (11): α_î(j) in equation (11) is l different random numbers.
[0108]
[0109] Next, the concealed input data generation processing unit 525 multiplies the temporary input data protection random number 537 by the input data protection random number of the i-th layer to calculate the temporary input data protection random number of the i-th layer (step S804).
[0110] Next, the anonymized input data generation processing unit 525 adds the input data of the i-th layer and the temporary input data protection random number of the i-th layer to calculate the anonymized input data of the i-th layer, outputs it to the normal area 310, stores it as anonymized input data 516, and terminates the processing (step S805).
[0111] For example, the concealment input data of the i-th layer is calculated using equation (12): x_i′ in equation (12) is an example of concealment input data, and represents a matrix of m rows and n columns.
[0112]
[0113] Furthermore, for example, the concealment input data of the i-th layer is calculated using equation (13): x_i′ in equation (13) is an example of concealment input data, and represents a matrix of m rows and n columns.
[0114]
[0115] 9 is a flowchart showing an example of parameter concealment influence removal processing. The parameter concealment influence removal processing unit 533 of the TEE trust region 38 first reads the concealment intermediate data 517 of the i-th layer, which is the layer in question (step S901). Next, the parameter concealment influence removal processing unit 533 reads the parameter-protected random number of the i-th layer, which is the layer in question, from the parameter-protected random number 528 (step S902). Next, the parameter concealment influence removal processing unit 533 divides the concealment intermediate data 517 of the i-th layer by the parameter-protected random number of the i-th layer to calculate the parameter concealment influence-removed intermediate data of the i-th layer and store it in the TEE trust region 38 (step S903).
[0116] For example, the parameter concealment influence-removed intermediate data of the i-th layer is calculated using equation (14). y_î~ in equation (14) is an example of the parameter concealment influence-removed intermediate data concealment input data of the i-th layer, and represents a matrix.
[0117]
[0118] 10 is a flowchart showing an example of input concealment influence removal processing. The input concealment influence removal processing unit 538 of the TEE trust region 38 first reads the parameter concealment influence removal intermediate data of the i-th layer, which is the layer in question (step S1001). Next, the input concealment influence removal processing unit 538 reads the temporary input data protected random number 537 (step S1002). Next, the input concealment influence removal processing unit 538 reads the input concealment influence removal parameters of the i-th layer, which is the layer in question, from the input concealment influence removal parameters 529 (step S1003).
[0119] Next, the input concealment influence removal processing unit 538 multiplies the temporary input data protection random number 537 by the input concealment influence removal parameter of the i-th layer to calculate the temporary input concealment influence removal parameter of the i-th layer (step S1004).
[0120] Next, the input concealment influence removal processing unit 538 outputs the result of subtracting the temporary input concealment influence removal parameter of the i-th layer from the concealment parameter influence removal intermediate data of the i-th layer as the output data of the i-th layer, and ends the processing (step S1005).
[0121] For example, the output data of the i-th layer is calculated using Equation (15). Equation (15) is an example of the i-th layer output data that is the execution result of the input concealment influence removal process, and y_i is expressed as a matrix of Equation (16).
[0122]
[0123]
[0124] Furthermore, for example, the output data of the i-th layer is calculated using equation (17). Equation (17) is an example of the output data of the i-th layer, and y_i is expressed by the matrix of equation (16).
[0125]
[0126] 11 is a dialog diagram showing an example of a machine learning parameter concealment setting information input screen. The CPU 51 of the parameter holder terminal 50 displays the machine learning parameter concealment setting information input screen 1100 on the display device 55 in accordance with the machine learning parameter concealment setting information input screen display program loaded into the memory 52.
[0127] The target parameter input section 1101 of the machine learning parameter anonymization setting information input screen 1100 is an input form in which the user inputs identification information (such as file names) of machine learning parameters to be transmitted to the confidential information processing server 30. The total processing layer number input section 1102 is an input form in which the user inputs the total number of target machine learning parameters, or in which the CPU 51 reads and inputs the number of layers from the machine learning parameters. The parameter anonymization necessity input section 1103 is an input form in which the user specifies whether or not the target machine learning parameters need to be anonymized for the confidential information processing server 30, or in which the CPU 51 reads and inputs the anonymization requirements from the target machine learning parameters.
[0128] The layer processing type input fields 1111 to 1114 of the machine learning parameter anonymization setting information input screen 1100 are input forms in which the user inputs the type of operation for each processing layer, or in which the CPU 51 reads and inputs the type of operation for each layer from the target machine learning parameters. The trust region anonymization processing availability input fields 1121 to 1124 are input forms in which the user inputs whether or not the operation for the layer can be processed in the TEE trust region 38, or in which the CPU 51 inputs based on the type of operation for each layer from the target machine learning parameters. The normal region anonymization processing availability input fields 1131 to 1134 are input forms in which the user inputs whether or not the operation for the layer can be performed while being anonymized in the normal region 310, or in which the CPU 51 inputs based on the type of operation for each layer from the target machine learning parameters. The i-th layer processing execution area input sections 1141 to 1144 are input forms in which the user inputs whether the calculation for that layer is to be executed in the normal area 310 or the TEE trust area 38, or in which the CPU 51 inputs the target machine learning parameters based on the type of calculation for each layer.
[0129] The input confirmation operation section 1105 of the machine learning parameter anonymization setting information input screen 1100 is an input button that the user presses using the input device 56, such as a mouse, when confirming the content of the input to each form. The input cancellation operation section 1106 is an input button that the user presses using the input device 56, such as a mouse, when canceling the content of the input to each form and canceling the input.
[0130] 12 is a table showing an example of the data structure of each layer processing concealment setting information. As an example of each layer processing concealment setting information 524 for the TEE trust region 38, each layer processing concealment setting information 524 storing information on concealment settings for processing each layer of the machine learning parameters transmitted by the parameter holder terminal 50 is shown in FIG.
[0131] Each layer processing concealment setting information 524 in FIG. 12 includes columns for layer number, layer processing type, whether or not concealment processing is possible in the trusted region, whether or not concealment processing is possible in the normal region, and the region in which layer i processing is to be performed. Specifically, for example, record 5243 in the third row indicates that the processing type of the third layer of multiple processing layers is a process that executes a linear operation called dense. Record 5243 also indicates that concealment processing is possible in the trusted region while protecting confidential parameter data. Record 5243 also indicates that concealment processing is possible in the normal region while protecting confidential parameter data, and that the region in which the user-specified third layer processing is to be performed is the normal region. Naturally, the layer processing concealment setting information 524 available in the confidential information processing system 1 according to this embodiment is not limited to the above-described setting information data, and any data structure, such as a database, can be used.
[0132] (3) Effects of the embodiment When the confidential information processing server 30 of the above-described embodiment receives machine learning parameters, which are confidential information, it performs parameter concealment processing of the machine learning parameters in the TEE trusted domain before placing the machine learning parameters in a normal domain where high-speed processing is possible, such as a GPU. This makes it possible to prevent information leakage of the machine learning parameters, which are confidential information, from a server administrator who can access data in the normal domain of the confidential information processing server 30 or a cyber attacker who has usurped the administrator's authority.
[0133] Furthermore, when the confidential information processing server 30 executes an inference process consisting of multiple layers of processing, such as a CNN, on confidential input information in the normal domain, it executes a concealed input data generation process in the TEE trust domain and then outputs the concealed input data to the normal domain. Then, in a normal domain capable of high-speed processing, such as a GPU, it executes processing of each layer, such as a CNN, based on the concealed input data and the concealed machine learning parameters to generate concealed intermediate data. Then, it inputs the generated concealed intermediate data into the TEE trust domain and executes a parameter concealment influence removal process and an input concealment influence removal process to calculate the inference process result. Then, it encrypts the inference process result, which is confidential information, in the TEE trust domain and transmits the process result to the user terminal 20 or the data holder terminal 10. Therefore, the confidential input data, machine learning parameters, and inference results can always be handled in a concealed or encrypted state when stored or processed in a normal domain capable of high-speed processing, such as a GPU. This prevents confidential information such as input data, machine learning parameters, and inference processing results from being leaked to attackers of the confidential information processing server 30 when inference processing consisting of multiple layers of processing is executed in a normal area where high-speed processing is possible, such as a GPU.
[0134] Furthermore, when the confidential information processing server 30 executes an inference process consisting of multiple layers of processing, such as a CNN, on confidential input information in the normal domain, it generates a temporary random number to be used only for that input information in the TEE trusted domain. Then, it executes a concealed input data generation process including a different random number for each piece of input information, and then outputs the concealed input data to the normal domain. Therefore, when a different temporary random number is not used for each piece of input information, a server administrator or a cyber attacker of the confidential information processing server 30 may intentionally set the input information to 0 to perform the concealed input data generation process. This makes it possible to identify the input data protection random number and prevent information leakage caused by exposing the input information.
[0135] Furthermore, the confidential information processing server 30 can specify whether to process each layer of the machine learning parameters in the normal region or the TEE trust region, based on the layer processing concealment setting information specified by the user of the parameter holder terminal 50. This makes it possible to specify the region in which to execute the processing for each layer, taking into consideration the feasibility of processing by a GPU that differs for each type of layer of the machine learning parameters, the presence or absence of a speed-up effect, and the like, and thereby makes it possible to improve the efficiency of inference processing consisting of processing for multiple layers, such as CNN.
[0136] (4) Other Embodiments (4-1) Executor of the i-th Layer Processing In the above-described embodiment, the secret information processing server 30 executes the i-th layer processing in the normal area 310, and performs concealment and de-concealment of the i-th layer parameters and the i-th layer input data in the TEE trusted area 38. The i-th layer processing does not necessarily have to be executed by the secret information processing server 30, but may be executed by an external device having a normal area that is connected to the secret information processing server 30 via communication. That is, the secret information processing server 30 outputs an instruction to execute the i-th layer processing to the external device. Then, in response to the execution instruction, the external device executes the i-th layer processing in the normal area and returns the execution result to the secret information processing server 30. Then, the secret information processing server 30 may obtain the execution result of the i-th layer processing from the external device.
[0137] (4-2) Temporary Random Number for Protecting Input Data in Layer i In the above-described embodiment, the input data in layer i is concealed using the layer i temporary input data protection random number α_i·r_i (formula (12)) obtained by multiplying the layer i input data protection random number (formula (5)) by the temporary random number α_i (formula (10)) that protects the input data in layer i. However, it is not necessarily necessary to use the temporary random number α_i in layer i, and "1" may be used instead of the temporary random number α_i in layer i.
[0138] (4-3) i-th Layer Input Data Protection Random Number In the above-described embodiment, the i-th layer input data is concealed using the i-th layer temporary input data protection random number α_i·r_i (equation (12)) obtained by multiplying the i-th layer input data protection random number (equation (5)) by the temporary random number α_i (equation (10)) that protects the i-th layer input data. However, it is not necessary to use the i-th layer input data protection random number; instead of the i-th layer input data protection random number, a matrix of the same size as the i-th layer input data protection random number and whose elements are all "1" may be used.
[0139] The present invention is not limited to the above-described embodiments, and includes various modifications and equivalent configurations within the spirit of the appended claims. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and the present invention is not necessarily limited to those including all of the described configurations. Furthermore, some of the configurations of a certain embodiment may be added to, deleted from, or replaced with other configurations.
[0140] Furthermore, the aforementioned configurations, functions, processing units, processing means, etc. may be partly or entirely realized in hardware, for example, by designing them as integrated circuits, etc. Furthermore, the aforementioned configurations, functions, processing units, processing means, etc. may be realized in software, by a processor interpreting and executing a program that realizes each function.
[0141] Information such as programs, tables, and files that realize each function can be stored in a storage device such as a memory, hard disk, or SSD (Solid State Drive), or in a recording medium such as an IC card, SD card, or DVD.
[0142] In addition, the control lines and information lines shown in the drawings are those that are considered necessary for explanation, and do not necessarily show all the control lines and information lines that are necessary for implementation. In reality, it can be considered that almost all components are interconnected.
[0143] In addition, the present invention can achieve the same effect when providing confidential information, such as trade secrets and other highly confidential information whose disclosure to outside the company is restricted by internal company rules, in addition to personal information, to external organizations, etc., by concealing the information.
[0144] Furthermore, the present invention can improve the processing efficiency of advanced deep learning and other technologies while keeping personal information and confidential business information secret in multiple organizations that handle personal information such as customer information and medical information, and confidential information such as trade secrets. As such, the present invention can be widely applied to information processing systems for public and private services that link a variety of confidential information.
[0145] 1: Secret information processing system, 30: Secret information processing server, 31: CPU, 38: TEE trusted region, 310: normal region, 312: i-th layer processing unit.
Claims
1. An information processing method performed by an information processing system that executes information processing using a neural network consisting of multiple i-th layers (i is a natural number between 1 and N), the information processing system having a processor and a memory, the memory including an isolated area that is a TEE (Trusted Execution Environment) isolated from a normal area, the isolated area holding: i-th layer parameters used in processing of the i-th layer; i-th layer parameter protection random numbers that are random numbers that protect the i-th layer parameters; i-th layer input data that is input to the i-th layer; and i-th layer input data protection random numbers that are random numbers that protect the i-th layer input data, the processor in the isolated area executes, for each i-th layer, a parameter concealment process that multiplies the i-th layer parameters by the i-th layer parameter protection random numbers to generate i-th layer concealment parameters that conceal the i-th layer parameters, and outputs the i-th layer concealment parameters to the normal area, executing, for each layer, an input concealment influence removal parameter generation process for generating an i-layer input data concealment removal parameter by multiplying the i-layer parameter by the i-layer input data protection random number to remove the concealment applied to the i-layer input data, and storing the generated parameter in the isolation area; executing an concealment input data generation process for adding the i-layer input data protection random number to the i-layer input data to generate i-layer concealed input data by concealing the i-layer input data, and outputting the i-layer concealed input data to the normal area as the i-layer input data; and executing a parameter concealment influence removal process for obtaining from the normal area an execution result of the i-layer processing executed on the i-layer concealed input data in the normal area based on the i-layer concealment parameter, and dividing the execution result by the i-layer parameter protection random number; executing an input concealment influence removal process in which the i-layer input data concealment removal parameter is subtracted from a result of the execution of the parameter concealment influence removal process, and setting the result of the input concealment influence removal process as i-layer output data output from the i-layer.
2. An information processing method as described in claim 1, wherein the memory includes the normal area, and the processor includes a process of executing, in the normal area, the i-th layer processing on the i-th layer concealed input data based on the i-th layer concealment parameters, and outputting the execution result of the i-th layer processing to the isolated area.
3. An information processing method as claimed in claim 1, characterized in that the processor, in the isolated area, executes the concealment input data generation process of generating a temporary random number for each of the layer i processing for each of the layer i input data, multiplying the layer i input data protection random number by the temporary random number and adding the result to the layer i input data to generate the layer i concealment input data, outputting the layer i concealment input data to the normal area as the layer i input data, and executing the input concealment influence removal process of subtracting the result of multiplying the layer i input data concealment removal parameter by the temporary random number from the result of execution of the parameter concealment influence removal process, and setting the result of execution of the input concealment influence removal process as the layer i output data output from the layer i.
4. An information processing method according to claim 3, characterized in that the processor uses, instead of the i-th layer input data protection random number, a matrix with all elements of 1 and of the same size as the i-th layer input data protection random number.
5. An information processing method as described in claim 1, characterized in that the isolated area holds each layer processing concealment setting information which is information indicating whether the processing of the i-th layer is to be performed in the normal area or the isolated area, and the processor executes the processing of the i-th layer in either the normal area or the isolated area according to the each layer processing concealment setting information.
6. An information processing method according to claim 1, wherein the neural network is a CNN (Convolutional Neural Network), and the i-th layer parameters are machine learning parameters of the CNN.
7. An information processing system that executes information processing using a neural network consisting of multiple i-th layers (i is a natural number between 1 and N), comprising a processor and a memory, the memory including an isolated area that is a TEE (Trusted Execution Environment) isolated from a normal area, the isolated area holding: i-th layer parameters used in processing of the i-th layer; i-th layer parameter protection random numbers that are random numbers that protect the i-th layer parameters; i-th layer input data that is input to the i-th layer; and i-th layer input data protection random numbers that are random numbers that protect the i-th layer input data, the processor in the isolated area executes, for each i-th layer, a parameter concealment process that multiplies the i-th layer parameters by the i-th layer parameter protection random numbers to generate i-th layer concealment parameters that conceal the i-th layer parameters, and outputs the i-th layer concealment parameters to the normal area, executing, for each layer, an input concealment influence removal parameter generation process for generating an i-layer input data concealment removal parameter by multiplying the i-layer parameter by the i-layer input data protection random number to remove the concealment applied to the i-layer input data, and storing the generated parameter in the isolation area; executing an concealment input data generation process for adding the i-layer input data protection random number to the i-layer input data to generate i-layer concealed input data by concealing the i-layer input data, and outputting the i-layer concealed input data to the normal area as the i-layer input data; and executing a parameter concealment influence removal process for obtaining from the normal area an execution result of the i-layer processing executed on the i-layer concealed input data in the normal area based on the i-layer concealment parameter, and dividing the execution result by the i-layer parameter protection random number; an input concealment influence removal process is executed in which the i-layer input data concealment removal parameter is subtracted from a result of the execution of the parameter concealment influence removal process, and the result of the input concealment influence removal process is set as i-layer output data output from the i-layer.
8. An information processing system as described in claim 7, wherein the memory includes the normal area, and the processor performs, in the normal area, the i-th layer processing on the i-th layer concealed input data based on the i-th layer concealment parameters, and outputs a result of the i-th layer processing to the isolation area.
9. An information processing system as described in claim 7, wherein the processor, in the isolated area, generates a temporary random number for each layer i processing for each layer i input data, executes the concealment input data generation process to generate the layer i concealment input data by adding a result of multiplying the layer i input data protection random number and the temporary random number to the layer i input data, and outputs the layer i concealment input data to the normal area as the layer i input data, executes the input concealment influence removal process to subtract a result of multiplying the layer i input data concealment removal parameter and the temporary random number from a result of execution of the parameter concealment influence removal process, and sets the result of execution of the input concealment influence removal process as the layer i output data output from the layer i.
10. An information processing system according to claim 9, wherein the processor uses, instead of the i-th layer input data protection random number, a matrix with all elements of 1 and of the same size as the i-th layer input data protection random number.
11. An information processing system as described in claim 7, characterized in that the isolated area holds each layer processing concealment setting information which is information indicating whether the processing of the i-th layer is to be performed in the normal area or the isolated area, and the processor executes the processing of the i-th layer in either the normal area or the isolated area according to the each layer processing concealment setting information.
12. An information processing system according to claim 7, wherein the neural network is a CNN (Convolutional Neural Network), and the i-th layer parameters are machine learning parameters of the CNN.
Citation Information
Patent Citations
Operation method and device of artificial intelligence AI model
CN110704850A
Commission of processing operation using homomorphism code
JP2019191575A
Information processing program and information processing device
JP2023027612A
Privacy-preserving machine learning
US20210112038A1
Information processing device, information processing method, and information processing program
WO2021140574A1
