Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Encryption and decryption method of deep learning model

A technology of deep learning and encryption methods, which is applied in the field of encryption and decryption of deep learning models, can solve problems such as lack of protection of assets, and achieve the effect of ensuring encryption security and preventing random theft

Pending Publication Date: 2022-07-05
深见清影科技(厦门)有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

As such, there is a lack of adequate protection for the designer's work product or the owner's assets

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Encryption and decryption method of deep learning model
  • Encryption and decryption method of deep learning model
  • Encryption and decryption method of deep learning model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0044] Embodiment 1: an encryption and decryption method for a deep learning model, the decryption module is divided into the following steps:

[0045] Step 1: Obtain the local machine code, and generate the corresponding ASCII code value sequence list;

[0046] Step 2: According to the number of encryption layers of the model, start the encryption layer, the ASCII code value and the pkl format file storing the real parameters of the model, and decrypt the pth format model file.

[0047] In the present invention, preferably, in step 1, the machine code is a unique identification code representing a computer device, which is composed of a series of letters, numbers and symbols. Therefore, first obtain the machine code of the model device to be deployed to facilitate subsequent encryption operations.

[0048] In the present invention, preferably, in step 2, the machine code is composed of letters, numbers and symbols. In order to facilitate the calculation, the corresponding ch...

Embodiment 2

[0056] Embodiment 2: a kind of encryption and decryption method of deep learning model, the decryption module is divided into the following steps:

[0057] Step 1: Obtain the local machine code, and generate the corresponding ASCII code value sequence list;

[0058] Step 2: According to the number of encryption layers of the model, start the encryption layer, the ASCII code value and the pkl format file storing the real parameters of the model, and decrypt the pth format model file.

[0059] In the present invention, preferably, in step 1, the machine code of the device is obtained, and each character of the machine code is converted into an ASCII code value, thereby obtaining a machine code sequence table.

[0060] In the present invention, preferably, in step 2, for the SL to SL+EL-piece layer parameters of the model file, cycle the machine code sequence table C, and according to the ASCII code value C in the list i Replace the encrypted value of the ith parameter of this l...

Embodiment 3

[0062] Embodiment 3: a kind of encryption and decryption method of deep learning model, the decryption module is divided into the following steps:

[0063] Step 1: Obtain the local machine code, and generate the corresponding ASCII code value sequence list;

[0064] Step 2: According to the number of encryption layers of the model, start the encryption layer, the ASCII code value and the pkl format file storing the real parameters of the model, and decrypt the pth format model file.

[0065] In the present invention, preferably, in step 1, the machine code is a unique identification code representing a computer device, which is composed of a series of letters, numbers and symbols. Therefore, first obtain the machine code of the model device to be deployed to facilitate subsequent encryption operations.

[0066] In the present invention, preferably, in step 2, the machine code is composed of letters, numbers and symbols. In order to facilitate the calculation, the correspondin...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a deep learning model encryption and decryption method, which comprises an encryption module and a decryption module, and is characterized in that the encryption module comprises the following steps: step 1, obtaining a machine code of a to-be-deployed model device; 2, generating an ASCII (American Standard Code for Information Interchange) code value sequence table of the machine code; 3, obtaining model parameters of the pth format file, and setting the number of encryption layers and a starting encryption layer according to the maximum number of layers of the model parameters; 4, according to the ASCII code value, the value of the corresponding position of each layer of the to-be-encrypted parameters is adjusted to be a random number, the number of encryption layers and the starting encryption layer of the model parameters are set, the multiple special positions of the multi-layer parameters of the model file are adjusted in combination with the equipment machine code, and due to the fact that the number of encryption layers is set, the number of the encryption layers is increased; the starting encryption layer and the machine code are different, the adjustment parameters and positions of the model file are different, and therefore the encryption safety of the model is guaranteed, and the model deployed on a client computer is effectively prevented from being stolen at will.

Description

technical field [0001] The invention belongs to the technical field of computer engineering, and in particular relates to an encryption and decryption method of a deep learning model. Background technique [0002] Deep learning (DL, Deep Learning) is a new research direction in the field of machine learning (ML, Machine Learning), which is introduced into machine learning to make it closer to the original goal - artificial intelligence (AI, Artificial Intelligence). Deep learning is to learn the inherent laws and representation levels of sample data, and the information obtained during these learning processes is of great help to the interpretation of data such as text, images and sounds. Its ultimate goal is to enable machines to have the ability to analyze and learn like humans, and to recognize data such as words, images, and sounds. Deep learning is a complex machine learning algorithm that has achieved results in speech and image recognition far exceeding previous rela...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06F21/60G06F40/126G06F40/157
CPCG06F21/602G06F40/126G06F40/157
Inventor 黄炜宸
Owner 深见清影科技(厦门)有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com