An integrated web service solution for industrial deep learning use cases

Abstract

The web application development landscape becomes capricious with the emergence of new technologies. The service providers attempt to integrate as much features on the web platform as they can. In the industry of machine learning and deep learning research, more and more companies prefer to convey big data analysis report via web application based on the various information visualization features. Web application offers convenience for industrial machine learning and deep learning researches. Web application could generate the real-time feedback to the client side in a seamless way. On the other hand, novel web application usually contains different modules and uses various types of communication protocol, which imposes great challenge over the maintenance job. Service orchestration and Continuous Integration and Continuous Deployment (CI/CD) could play an important role in dealing with modern software architecture.

This study investigates how progressive web apps and corresponding back-end system can be designed, implemented and integrated together into a workflow. In addition, this study takes a further look at an efficient way to communicate between different web services. In the range of the project, JavaScript serves as the prioritized programming language for developing the web service and Kubernetes as the fundamental cluster orchestration tool. On top of that, the compatibility research and performance evaluation on some HTTP/2.0 features is included in the project.