Motivation: 

In this talk, we will discuss leveraging Machine Learning practices in Software Testing with several practical examples and a case study that I used in my project to do Bug Triage. Let's embrace the future together!

Problems: 

Testing is very cumbersome. Agile is open to changes, which means expected behaviors can change over time. Besides, due to implementation changes, tests may be broken. And most importantly, time is very precious and limited. Manual efforts should be minimized to improve coverage and reserve more time for exploratory activities with limited resources.

Solutions: 

Manual effort can be reduced and testing can be done in a more convenient and consistent way by applying ML. Stages in which ML is applicable are:

• Test definition
• Automatic code generation
• Execution: exploratory testing.
• Maintenance and grouping,
• Review test code.

Results & Conclusion: 

We see how ML helps in all stages. I summarize the application areas with algorithms and discuss AI applications' advantages and potential risks in software testing.

To sum up, this talk targets an important problem, AI-based applications of software testing. AI is one of the hottest topics in the software world nowadays. Especially, mining valuable information from bugs can be made use of by managers to guide feature priorities. I introduce the applications in different testing stages, making it easy for the audience to find what they want.

Outline

• Introduction 
• ML Practices for Test Case Definition
• ML Practices for Test Implementation
• ML Practices for Test Execution
• ML Practices for Test Maintenance: Self-Healing & Refactoring
• Case-study: Estimation of severity of upcoming bugs according to trained model 
• Close & Questions