Wang Receives MA ACORN Innovation Award
Electrical and Computer Engineering Assistant Professor Yanzhi Wang recently received a 2020 ACORN Innovation Award from The Massachusetts Technology Transfer Center for his project entitled “Real-Time Acceleration of Deep Learning Applications on Mobile Edge Devices”.
Deep learning or deep neural networks (DNNs) have become the fundamental element and core enabler of ubiquitous artificial intelligence. The current DNN execution frameworks on mobile edge devices support only a small subset of DNNs, and the performance is far from real time with many important types of DNNs, like recurrent neural networks useful in natural language processing and video understanding, not supported. These limitations are overcome with the development of an acceleration framework of deep learning applications, which can achieve real-time performance using currently available mobile and edge devices. This crosslayer optimization framework has full vertical integration of algorithm, compiler, and hardware levels. The Acorn funds will be utilized for making demonstrations on a smartphone for four key applications (and DNNs): speech recognition, object detection in video input, human activity detection, and natural language processing, key to a successful path to commercialization.
The MA Acorn Innovation Fund is intended to support the demonstration of the viability of a technology developed at Massachusetts research universities. Awards could be used to further develop a prototype or to gather additional data to demonstrate proof of concept or to obtain data that shows how the technology compares to existing technologies and what the competitive advantages are.