Mohak Chadha’s proposal on “FedLessExit: Leveraging Early Exiting for Client Model Personalization in Serverless Federated Learning” has been accepted at the Software Campus partnered with Huawei Munich research centre and will receive a funding of 100,000 euros.
About Software Campus:
Software Campus aims at combining industry and science together in order to train the IT managers of the future. It specifically combines cutting-edge research and practical relevance. The software campus is funded by the Federal Ministry of Education and Research (BMBF). Industrial companies DATEV eG, Holtzbrinck Publishing Group, Huawei, IAV GmbH, Merck, Rohde & Schwarz, Software AG, TRUMPF Group, Carl Zeiss AG are partners of this program.
About Proposal:
Federated Learning (FL), a new paradigm of distributed ML training, enables the collaborative training of ML models by clients and addresses the problems of privacy of data. However, training a global model via FL leads to several challenges such as resource inefficiency, high costs, and low accuracy of the trained global model when inferenced by the client. This project focuses on mitigating these challenges by utilizing Serverless Computing technologies in combination with Early Exiting.
Mohak Chadha selected as a participant of the Software Campus Program with a funding of 100.000 Euros from BMBF