AI4Perf - Artificial Intelligence based Modeling and Automation of Networked Systems

The project “AI4Performance” aims for developing a smart approach for performing tests of networked systems and evaluating them automatically, thereby collecting meaningful data sets, and processing them using advanced methods including an innovative machine learning method, Graph Neural Network. Use cases of the two industrial partners Cisco and Airbus are representative for a larger range of use cases in the industry in which performance issues of complex networked systems are very difficult to be solved by today’s methods.

Goal

The project aims for answering the following questions: What-if analysis for predicting the impact of changes on performance, root-cause analysis and bottleneck discovery, and detection of performance anomalies. The project aims to develop machine learning based methods to answer these questions, which will have a wide applicability in many industrial use cases, and as a consequence having a large industrial and societal impact.