Automated Testing with Machine Learning Frameworks: A Critical Analysis †

Abstract

As software systems are becoming more and more complex and standard testing practices are exhausting, we need smart solutions to reduce the time, efforts and resources spent on software testing. The aim of this paper was to critically analyze machine learning (ML) frameworks related to software automation. We measured the performance of testing tools on the basis of the manual labor (effort) required, in addition to the test performance, accuracy or error rate, scope, time required and prerequisite knowledge requirements. These factors play a vital role to ensure ML frameworks with automation software can produce great results and hence improve software quality.