Serverless Computing Approach for Deploying Machine Learning Applications in Edge Layer
Ta Phuong Bac, Minh Ngoc Tran, Younghan Kim
Abstract
Serverless computing-a stateless cloud computing model, is an emerging solution that has shown significant benefits to efficiency and cost for event-driven applications in the cloud environment, including artificial intelligence (AI), machine learning applications. With serverless computing, the machine learning system’s complexity is minimized, flexible and straightforward in management. However, operating and managing serverless machine learning services on clouds faces many limitations such as latency and data privacy. Local distributed edge computing nodes which are closed to users can address these challenges of cloud-serverless AI applications. Based on this motivation, in this paper, we propose an architecture for deploying machine learning workload as serverless functions in the edge environment. We illustrate our proposed approach and evaluate its performance and effectiveness by exploiting a holistic end-to-end image classifier, a famous machine learning use case in the MNIST dataset. Our proof of concept provides comprehensive assessments that prove its effectiveness in latency reduction and distributed machine learning deployment.