Challenges and opportunities in deploying machine learning systems on edge devices
Halls department, Hall 4
Thursday, 27 December 2018
15:15 - 16:15
Edge computing enables advanced on-device processing and decision making. This paradigm lowers dependence on the cloud and hence reduces the latency for critical applications and potentially enhances the privacy of users. On the other hand, it entails certain design and implementation challenges, including but not limited to, mapping computation and/or communication intensive algorithms to resource-constrained devices, providing essential portability and accessibility features, and ensuring security and safety at the edge. In this talk, we will go over some of these challenges, particularly in the scope of deploying machine learning systems on edge devices. Furthermore, we will talk about some of the proposed solutions and the opportunity landscape for edge computing.
Majid Sabbagh graduated from Isfahan University of Technology in 2013 with B.Sc. degree in electrical engineering-electronics. As the final project, he designed and implemented a full E1 transceiver targeting off the shelf Field Programmable Arrays (FPGAs) supervised by Prof. Hossein Saidi and Prof. Ali Ghiasian. Toward the end of his undergraduate studies, he joined SarvNet at Isfahan Science and Technology Town in which he focused on studying modern switches and routers. In spring 2014, he started his graduate studies in computer engineering at Northeastern University; doing research on embedded computer vision and medical signal processing. He earned an M.Sc. degree with a thesis on accelerating cardiac MRI compressed sensing image reconstruction using Graphics Processing Units (GPUs) supervised by Prof. Miriam Leeser and Prof. Mehdi H. Moghari. Currently, he is pursuing his Ph.D. at Northeastern University Energy-Efficient and Secure Systems Lab (NUEESS) under the supervision of Prof. Yunsi Fei. He contributed to various applied projects as well as conference and journal articles.