Toward Open-Source Cloud-Based Visual Machine Learning Platform: A Human-Interface Usability Study
Academic Article
Overview
Overview
Abstract
The widespread adoption of AI in almost all aspects of our daily lives dictates a need for a significant shift in the Machine Learning (ML) teaching and development paradigms where problem-domain experts should take matters into their own hands to design, train, test, and deploy more specialized and efficient ML models. However, given the complexity of its concepts, teaching machine learning in general and neural networks in particular remains a challenging. Thus, shielding problem-domain experts with interactive learning environments can reduce the cognitive load needed to grasp ML low-level details, which is of paramount importance to achieving this goal. This study is the first step to empower problem-domain experts with an open-source cloud-based machine learning platform where they can create and use their own ML models while not being concerned with the software development technicalities. To investigate the usability and user-friendliness of the user-interface of such system, a prototype was built for an experiment. The prototype was introduced to participants from different backgrounds and their feedback were captured.
