Data Augmentation Algorithm For IIoT Federated Learning:Design And Implement

Nowadays,machine learning is widely used in our daily life,and it plays an important role in the industrial field.However,in the industrial Internet of Things environment,the data on many front-end devices involves the issue of privacy protection.As a distributed machine learning model,federated learning can directly perform data training on the client,and can make full use of the data.At the same time,the FL ensures the privacy and security of the client,and successfully solves the privacy protection problem in the Industrial Internet of Things scenario.However,in the industrial Internet of Things scenario,the data collected by the front-end equipment has a large difference,which belongs to non-independent and identically distributed data,and non-independent and identically distributed data has a greater impact on the performance of the federated learning.This paper well use Generative Adversarial Networks to achieve data enhancement,and then reduce the impact the federated learning model,the main work of this paper is as follows:Firstly,analyze the performance of the IIoT federated learning model,and verify the analysis results through simulation experiments.This paper first briefly explains the reason why the data set obtained by the front-end equipment will conform to Non-IID in the industrial Internet of Things environment,and then introduces the impact of the Non-IID data set on the performance of the federated learning algorithm,and then verifies it through simulation experiments.Seconely,using Generative Adversarial Networks to realize the generation of fake data,and realize data enhancement on front-end devices.Use the GAN to generate the feature type fake data that the front-end equipment lacks,and use the generated data to participate in the subsequent federated learning training to enhance the performance of the federated learning model.In order to reduce the communication overhead while ensuring the performance of the GAN model,we propose an adaptive front-end device selection algorithm.In order to ensure that the data transmission process is not subject to malicious attacks,we propose a picture perturbation encryption algorithm.And in order to improve the utilization of generated data,this paper proposes an adaptive data generation algorithm.Thirdly,validation of Algorithms by Simulation Experiments.Using the MNIST dataset and CIFAR-10 dataset,the Dirichlet distribution is used for data sampling to generate non-independent and identically distributed datasets,and the Fed Avg model is used as a reference to verify the effectiveness of the data enhancement algorithm proposed in this paper.The results show that the data enhancement algorithm proposed in this paper can effectively reduce the energy consumption of the front-end devices and greatly improve the accuracy of the model and communication efficiency.