Method and system for implementing binary convolutional neural network based on homomorphic encryption

The invention discloses a binary convolutional neural network implementation method and system based on homomorphic encryption, and the method comprises the steps: carrying out the preprocessing of an MNIST data set through a cloud end, and obtaining binary image data; the cloud constructs a binary convolutional neural network model under a plaintext, and network parameters are trained; the client performs homomorphic encryption on the binary image data by using a TFHE encryption scheme, and transmits the encrypted image data to the cloud; the cloud end constructs an encryption neural network based on a Boolean circuit or an arithmetic circuit according to the network parameters obtained through training, carries out reasoning on the encrypted image data to obtain a secret state result, and transmits the secret state result to the client; and the client decrypts the secret state result to obtain a reasoning result. According to the method, the TFHE homomorphic encryption algorithm is combined with the binary