首页|Reports from PSG College of Technology Advance Knowledge in Machine Learning (Fe dassess: Analysis for Efficient Communication and Security Algorithms Over Vario us Federated Learning Frameworks and Mitigation of Label-flipping Attack)
Reports from PSG College of Technology Advance Knowledge in Machine Learning (Fe dassess: Analysis for Efficient Communication and Security Algorithms Over Vario us Federated Learning Frameworks and Mitigation of Label-flipping Attack)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News-Current study results on Machine Learning have be en published. According to news reporting out of Tamil Nadu, India, by NewsRx ed itors, research stated, "Federated learning is an upcoming concept used widely i n distributed machine learning. Federated learning (FL) allows a large number of users to learn a single machine learning model together while the training data is stored on individual user devices." Our news journalists obtained a quote from the research from the PSG College of Technology, "Nonetheless, federated learning lessens threats to data privacy. Ba sed on iterative model averaging, our study suggests a feasible technique for th e federated learning of deep networks with improved security and privacy. We als o undertake athorough empirical evaluation while taking various FL frameworks a nd averaging algorithms into consideration. Secure multi party computation, secu re aggregation, and differential privacy are implemented to improve the security and privacy in a federated learning environment. In spite of advancements, conc erns over privacy remain in FL, as the weights or parameters of atrained model may reveal private information about the data used for training. Our work demons trates that FL can be prone to label-flipping attack and a novel method to preve nt label-flipping attack has been proposed. We compare standard federated model aggregation and optimization methods, FedAvg and FedProx using benchmark data se ts. Experiments are implemented in two different FL frameworks-Flower and PySy ft and the results are analyzed. Our experiments confirm that classification acc uracy increases in FL framework over a centralized model and the model performan ce is better after adding all the security and privacy algorithms."
Tamil NaduIndiaAsiaAlgorithmsCyb ersecurityCyborgsEmerging TechnologiesMachine LearningPSG College of Tec hnology
NETL
NSTL
