Research on Neural Network Classification Prediction Based on Fully Homomorphic Encryption in Cloud Computing Environment
In the context of cloud computing,the data sharing and service outsourcing of data mining applications have generated enormous wealth,but they have also brought risks of privacy leaks,and their information security problems need to be urgently addressed.This article analyzes the risks existing in neural network prediction classification service outsourcing and the application of fully homomorphic encryption algorithms in the cloud computing environment.It uses the square function as the nonlinear activation function in the neural network,designs and implements a ciphertext-based data classification prediction solution,and provides a detailed description of the five main steps of the solution operation:model training,model encryption,data encryption,ciphertext computation,and decryption of results.The designed test experiment results verify the feasibility of the solution,ensuring that the model and data are not leaked during the entire neural network processing process,effectively guaranteeing the security and privacy of the data and model.
万方数据
维普
