Explore the intersection of software testing and machine learning in this 46-minute conference talk. Delve into the challenges of modern software testing, particularly in GUI and complex system testing, and examine how machine learning techniques can potentially address these issues. Learn about GUI regression testing, test abstractions, and the application of supervised and unsupervised learning in testing scenarios. Discover the potential of reinforcement learning for testing games like Candy Crush, and gain insights into fuzzing and testing machine learning systems themselves. Evaluate the promising aspects and limitations of applying machine learning to software testing, and understand how these advancements may shape the future of quality assurance in software development.
Overview
Syllabus
Intro
What is software testing?
The problems with testing today
The promise of Machine Learning for testing
Problem: GUI testing is a mess
GUI Regression testing
Test abstractions in GUI-testing
Example: abstract GUI testing
GUI-abstraction with Supervised Learning
The ingredients of a clustering solution
The bag-of-words model
Characteristics of Unsupervised Learning
Unsupervised Learning: Criteria for success
Testing Candy Crush with Reinforcement Learning
How to succeed with Reinforcement Leaming
Related: Fuzzing
Testing machine learning systems
Taught by
NDC Conferences
