The work involves overcoming communication overhead, often a bottleneck in distributed learning and implementations of deep models. Advancement in this space would allow companies to scale up the number of workers leading to accelerated training and computations. Fekri's focus is on the challenges encountered due to transferring data and information (such as parameters of the model) among workers or between workers and a central node. During this project, Fekri will introduce solutions to overcome these issues.

“I am honored to receive this grant and am looking forward to the progress that will be made with Sony’s support. My hope is that this work will lead to more efficient communication between distributed learning and the implementation of deep learning models,” said Fekri.