Deep Resilience for Multifaceted Federated Learning in Internet-of-Everything

Principal Investigator:Jin-Hee Cho, associate professor, computer science; director, Trustworthy Cyberspace Lab, Virginia Tech

Co-Principal Investigators: 
Hoda Eldardiry, associate professor, computer science; director, Machine Learning Lab, Virginia Tech

Hongyi Wu, Batten Chair Professor, director, School of Cybersecurity, electrical and computer engineering, Old Dominion University

Rui Ning, research assistant professor, The Center for Cybersecurity Education and Research, Old Dominion University

Deep Resilience for Multifaceted Federated Learning in Internet-of-Everything Project Description: 

Federated learning (FL) is mainly involved with training statistical models in remote devices (e.g., mobile phones) or isolated data centers (e.g., hospitals or troops) where data are localized. Training in heterogeneous and potentially large networks brings new challenges in dealing with large-scale machine learning, distributed optimization, and privacy-preserving data analysis. As local devices’ storage and computational capabilities grow fast, leveraging the enhanced local resources on each device has been a possible solution. Further, to avoid transmit raw data by the remote devices, FL has been considered as a promising solution by directly training statistical models on the remote devices. Particularly, the research team will consider an Internet-of-Everything (IoE) as a network environment for FL to be deployed where IoE devices are highly heterogeneous in their capabilities, data modalities, and services.