Encryption method of computer system

Abstract

The invention discloses an encryption method of a computer system, and the computer system comprises the following components: a, an encryption module, which is used for encrypting input data; the decryption module b is used for decrypting the encrypted data; the secret key management module is used for generating, storing and managing secret keys required by encryption and decryption; and d, a deep learning model module used for training and applying a deep learning model to improve the encryption algorithm. Compared with the prior art, the computer system encryption method based on deep learning has the advantages that the technologies of the deep convolutional neural network, the pre-trained residual network, the variational auto-encoder and the like are utilized; the security, efficiency, flexibility and expansibility of an encryption algorithm are improved, and an encryption system can better cope with current complex security challenges.

Description

Technical Field

[0001] The present invention relates to the technical field of computer encryption, and in particular to an encryption method for a computer system. Background Art

[0002] Traditional computer system encryption methods usually use traditional encryption algorithms, such as DES (Data Encryption Standard) and RSA (Rivest-Shamir-Adleman). These algorithms are based on mathematical principles and number theory, and use operations and specific keys to convert data into ciphertext to protect data security. However, with the improvement of computing power and the development of attack technology, the security of traditional encryption algorithms is increasingly threatened.

[0003] In current computer systems, with the rapid development of deep learning technology, people have begun to explore the application of deep learning in the field of encryption. Deep learning is a branch of machine learning that can extract features and patterns from data and make predictions and inferences by building multi-layer neural networks and using large-scale data for training and learning. Deep learning has achieved remarkable results in the fields of image recognition, speech recognition, and natural language processing. The present invention proposes an encryption method that combines deep learning with computer system encryption to cope with the current complex security challenges. Summary of the invention

[0004] In order to solve the above-mentioned various problems, the present invention provides an encryption method for a computer system.

[0005] To solve the above technical problems, the technical solution of the present invention is: an encryption method for a computer system, the computer system comprising the following components: a. Encryption module, used to encrypt input data. The encryption module uses a deep convolutional neural network to convert the input data into ciphertext through a series of convolutional layers, pooling layers, and fully connected layers; b. Decryption module, used to decrypt the encrypted data. The decryption module uses an inverse deep convolutional neural network to restore the ciphertext data to the original plaintext data through reverse convolution and pooling operations; c. A key management module, which is used to generate, store and manage the keys required for encryption and decryption. The key management module uses a variational autoencoder to generate a random symmetric encryption key and uses a hash function to map the key to a key of fixed length; d. Deep learning model module, used to train and apply deep learning models to improve encryption algorithms. The deep learning model module uses a pre-trained residual network for training and improves the security and efficiency of encryption algorithms by learning the distribution and characteristics of data.

[0006] Furthermore, the deep convolutional neural network adopts multiple layers of convolutional layers and pooling layers to extract spatial and temporal features of input data. Each convolutional layer contains multiple filters, which extract features of the input data through convolution operations, and then reduce the data dimension through pooling operations and enhance the invariance of features.

[0007] Furthermore, the decryption module adopts an inverse deep convolutional neural network to reversely transform the ciphertext data generated by the encryption module. The decryption module restores the original plaintext data in the encryption process layer by layer through inverse convolution and pooling operations.

[0008] Furthermore, the variational autoencoder can generate a random symmetric encryption key. The variational autoencoder uses an encoder to map input data to a probability distribution in a latent space, and uses a decoder to map the sampling points in the latent space back to the original data space to generate a symmetric encryption key.

[0009] Furthermore, the pre-trained residual network is used to train a deep learning model module to improve the performance and security of the encryption algorithm. The pre-training process uses a large-scale data set for training and adjusts the model parameters by minimizing the error between the predicted output and the actual output.

[0010] Furthermore, the key management module uses a hash function to map the generated symmetric encryption key into a key of fixed length. The hash function can map a key of any length into a byte sequence of fixed length for use by the encryption and decryption modules.

[0011] Furthermore, the deep learning model module uses an adaptive learning rate and batch normalization optimization algorithm to train the deep learning model. The adaptive learning rate dynamically adjusts the learning rate according to the training process of the model to speed up the convergence speed and improve the stability of the model. Batch normalization standardizes the input data during the training process to solve the problems of gradient disappearance and gradient explosion.

[0012] Furthermore, the encryption module and decryption module adopt a deep residual network structure. The deep residual network can effectively train the deep network by introducing jump connections, avoid the gradient disappearance problem, and improve encryption and decryption performance.

[0013] Furthermore, the encryption module and the decryption module adopt parallel computing to increase the speed of encryption and decryption. Parallel computing utilizes multiple processors or graphics processors to process data in parallel to speed up the execution speed of the encryption and decryption process.

[0014] Compared with the prior art, the present invention has the following advantages: 1) In the field of encryption, deep learning can be applied to the encryption and decryption process of data. By using deep convolutional neural networks instead of traditional encryption algorithms, data security can be better protected. Deep convolutional neural networks can convert input data into ciphertext through a series of convolutional layers, pooling layers, and fully connected layers, and reversely restore ciphertext data to original plaintext data. At the same time, pre-trained residual networks can improve the performance and security of encryption algorithms, and improve the encryption and decryption effects by learning the distribution and characteristics of data.

[0015] 2) In addition to deep convolutional neural networks, variational autoencoders are also a commonly used technology for generating random symmetric encryption keys. Variational autoencoders use encoders to map input data into probability distributions in latent space, and use decoders to map sampling points in the latent space back to the original data space to generate symmetric encryption keys. This method increases the randomness and unpredictability of the key, and improves the flexibility and anti-attack capabilities of the encryption algorithm.

[0016] 3) In addition, the key management module plays an important role in the encryption system. The module is responsible for generating, storing and managing the keys required for encryption and decryption. In the present invention, the key management module uses a hash function to map the generated symmetric encryption key to a key of fixed length. The hash function can map a key of any length to a byte sequence of fixed length for use by the encryption and decryption modules.

[0017] 4) In summary, the computer system encryption method based on deep learning utilizes technologies such as deep convolutional neural networks, pre-trained residual networks, and variational autoencoders to improve the security, efficiency, flexibility, and scalability of encryption algorithms, enabling encryption systems to better cope with current complex security challenges. Implementation

[0018] The present invention is further described in detail below in conjunction with the embodiments.

[0019] An encryption method for a computer system, comprising the following modules: Encryption Module: The encryption module uses a deep convolutional neural network (DCNN) to convert input data into ciphertext. The network includes multiple convolutional layers, pooling layers, and fully connected layers. Each convolutional layer contains multiple filters, which extract the spatial and temporal features of the input data through convolution operations. Then, the data dimension is reduced through pooling operations and the invariance of the features is enhanced. Finally, the features are converted into ciphertext through a fully connected layer.

[0020] Decryption Module: The decryption module uses a reverse deep convolutional neural network to reversely transform the ciphertext data generated by the encryption module. This module restores the original plaintext data in the encryption process layer by layer through reverse convolution and pooling operations.

[0021] Key Management Module: The key management module uses a variational autoencoder (VAE) to generate random symmetric encryption keys to improve the security of the encryption algorithm. VAE uses an encoder to map the input data to a probability distribution in the latent space, and uses a decoder to map the sampling points in the latent space back to the original data space to generate a symmetric encryption key. Subsequently, a hash function is used to map the generated symmetric encryption key to a key of fixed length.

[0022] Deep Learning Model Module: The deep learning model module uses a pre-trained residual network to train and improve the performance and security of the encryption algorithm. The pre-training process uses a large-scale data set to adjust the model parameters by minimizing the error between the predicted output and the true output. At the same time, it uses an adaptive learning rate and batch normalization optimization algorithm to improve the training effect and model stability.

[0023] Parallel Computing: In order to increase the speed of encryption and decryption, the encryption module and decryption module use parallel computing, using multiple processors or graphics processors to process data in parallel, which can speed up the execution of encryption and decryption.

[0024] Deep Residual Network: The encryption module and decryption module adopt a deep residual network structure and introduce skip connections to train the deep network. Skip connections can avoid the gradient vanishing problem and improve the performance of encryption and decryption.

[0025] System Architecture: The encryption module, decryption module, key management module and deep learning model module described above can form a computer system encryption method and system based on deep learning and variational autoencoder. The system can run in a computer system to encrypt and decrypt input data to protect data security.

[0026] The embodiments of the present invention can improve the encryption efficiency and security of computer systems by using technical means such as deep convolutional neural networks, reverse deep convolutional neural networks, variational autoencoders, and pre-trained residual networks. At the same time, through parallel computing and deep residual network structures, the speed of encryption and decryption is accelerated, and the gradient disappearance problem is avoided. This computer system encryption method can be widely used in various fields to protect the security of confidential data.

[0027] The present invention and its embodiments are described above, and such description is not restrictive. What is shown in the embodiment is only one of the embodiments of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A computer system encryption method, characterized in that: The computer system includes the following components: a. Encryption module, used to encrypt input data. The encryption module uses a deep convolutional neural network to convert the input data into ciphertext through a series of convolutional layers, pooling layers, and fully connected layers; b. Decryption module, used to decrypt the encrypted data. The decryption module uses an inverse deep convolutional neural network to restore the ciphertext data to the original plaintext data through reverse convolution and pooling operations; c. A key management module, which is used to generate, store and manage the keys required for encryption and decryption. The key management module uses a variational autoencoder to generate a random symmetric encryption key and uses a hash function to map the key to a key of fixed length; d. Deep learning model module, used to train and apply deep learning models to improve encryption algorithms. The deep learning model module uses a pre-trained residual network for training and improves the security and efficiency of encryption algorithms by learning the distribution and characteristics of data.

2. The encryption method of a computer system according to claim 1, characterized in that: The deep convolutional neural network uses multiple layers of convolutional layers and pooling layers to extract spatial and temporal features of input data. Each convolutional layer contains multiple filters. The input data is subjected to feature extraction through convolution operations, and then the data dimension is reduced through pooling operations, and the invariance of the features is enhanced.

3. The encryption method of a computer system according to claim 1, characterized in that: The decryption module uses a reverse deep convolutional neural network to reversely convert the ciphertext data generated by the encryption module. The decryption module restores the original plaintext data in the encryption process layer by layer through reverse convolution and pooling operations.

4. The encryption method of a computer system according to claim 1, characterized in that: The variational autoencoder can generate a random symmetric encryption key. The variational autoencoder uses an encoder to map input data to a probability distribution in a latent space, and uses a decoder to map the sampling points in the latent space back to the original data space to generate a symmetric encryption key.

5. The encryption method of a computer system according to claim 1, characterized in that: The pre-trained residual network is used to train a deep learning model module to improve the performance and security of the encryption algorithm. The pre-training process uses a large-scale data set for training and adjusts the model parameters by minimizing the error between the predicted output and the actual output.

6. The encryption method of a computer system according to claim 1, characterized in that: The key management module uses a hash function to map the generated symmetric encryption key into a key of fixed length. The hash function can map a key of any length into a byte sequence of fixed length for use by the encryption and decryption modules.

7. The encryption method of a computer system according to claim 1, characterized in that: The deep learning model module uses adaptive learning rate and batch normalization optimization algorithm to train the deep learning model. The adaptive learning rate dynamically adjusts the learning rate according to the training process of the model to accelerate the convergence speed and improve the stability of the model. Batch normalization standardizes the input data during the training process to solve the problems of gradient disappearance and gradient explosion.

8. The encryption method of a computer system according to claim 1, characterized in that: The encryption module and decryption module adopt a deep residual network structure. The deep residual network can effectively train the deep network by introducing jump connections, avoid the gradient disappearance problem, and improve the encryption and decryption performance.

9. The encryption method of a computer system according to claim 1, characterized in that: The encryption module and the decryption module adopt parallel computing to improve the speed of encryption and decryption. Parallel computing uses multiple processors or graphics processors to process data in parallel to speed up the execution speed of the encryption and decryption process.