T7: Introduction to Testing and AI


Date
July 24 (Wednesday)
Period
Half-day

Abstract


Artificial Intelligence (AI) is gaining momentum and is being implemented across diverse business as the digital transformation is demanding the adoption of new and emerging technologies. AI helps systems to perform tasks that would traditionally need human intellect. A computer can be fed with huge amount of data sets, which then adds logic and patterns to come up with relevant inferences. Smart Testing is needed to ensure that the results derived are relevant and in line with the business objectives.

Artificial Intelligence can and must be tested. The challenge in this case is in fact that a learning system will change its behavior over time. Predicting the outcome isn't easy because what's correct today may be different from the outcome of tomorrow that is also correct. But Artificial Intelligence can also be used to make testing smart, i.e. more effective and/or efficient. Black-box software testing has become a dominant part of quality assurance for industrial products when the system behavior is too complex to be represented by reusable design artifacts for testing.

This tutorial introduces test approaches for AI based systems as well as AI-based testing techniques. For their understanding a quick introduction into NNs in general is given and practical experience provided using one of the most popular open-source machine learning frameworks. The tutorial aims to deepen the understanding of NNs and how they can be tested and applied for testing.

Speaker


Christof J. Budnik avatar
Christof J. Budnik USA

Siemens Corporate Technology, USA


Dr. Christof J. Budnik is a Senior Key Expert Engineer at the Architecture and Verification of Intelligent Systems Research Group of Siemens Corporate Technology in Princeton, NJ. Dr. Budnik manages, leads, and advises business projects in several industrial domains such as rail solutions, building technologies, energy management, and healthcare applying advanced verification and validation technologies. His research focuses on the verification and validation of intelligent cyber-physical systems. His area of interest comprises the application of model-based testing, mutation testing, test automation, formal verification, and machine learning. He is author of over sixty publications at national and international journals and conferences which have been cited over four hundred times. Dr. Budnik is steering committee member of the international workshop on Automation of Software Test (AST), and has been program chair of software engineering conferences and workshops. He is a regular member of many program and organizing committees including the International Conference on Secure Software Integration and Reliability Improvement, International Conference on Software Engineering and Formal Methods, and International Conference on Software Security and Reliability. He further serves as guest editor and reviewer for selected journals including the Journal of Systems and Software, Journal of Software Testing, Verification & Reliability, and IEEE Transactions on Reliability.