Efficient Machine Learning in a Collaborative Edge-cloud Architecture

With the growing application of learning-based technology in image recognition, speech translation, and other back-end systems like ad advertising, artificial intelligence (AI) has finally moved from research labs to real business. As AI moves closer to reality, it is expected to help us more in more personalized and even critical contexts, like smart home, autonomous car. Traditional learning approaches start to meet new challenges in data collection, resource assumption, latency, since these contexts usually request near real-time interaction and high accuracy to guarantee safety and quality of service, or demand data that may be deemed as private, mostly even in resource-constrained environments. The objective of the proposed research is to examine and eventually improve the performance of artificial intelligence, especially deep learning approaches, in the edge-cloud architecuture. We will investigate the trade-offs among resources, latency, accuracy during the process of applying learning approaches from a system’s perspective. The proposed research can greatly accelerate the adoption of artificial intelligence into real-world production.

Faculty Supervisor:

Jiangchuan Liu

Student:

Partner:

The Hong Kong Polytechnic University

Discipline:

Computer science

Sector:

Information and Communications Technology; Technology; Transportation (excluding aerospace)

University:

Simon Fraser University

Program: