If programmed appropriately, AI systems are capable of learning solution behavior even from huge data sets. However, they must be trained to do so. For the development and application of AI-supported security systems, such as systems for detecting attacks against computer systems or computer networks, current, representative and extremely detailed training data are therefore essential. However, these are often not available due to data protection issues. To obtain training data that is as real as possible, imbus AG therefore launched the GENESIS research project together with Applied Security GmbH and the Coburg University of Applied Sciences. GENESIS is funded by the Bavarian Collaborative Research Program (BayVFP) of the Free State of Bavaria and started in April 2022 with an expected end in March 2023.
GENESIS stands for GENeration and Evaluation of Realistic Streams of Information for Training and Testing of Intelligent Security Systems.
The goal of GENESIS is to address this shortcoming in cybersecurity by developing methods for generating and evaluating realistic training and test data sets. Different security-relevant data sources such as network data or host-based log files are taken into account.Another goal of GENESIS is to develop a better understanding of how certain attack scenarios are reflected in different data sources. Generated training data, methods, and lessons learned will be exposed to enable more comprehensive crowd-based training opportunities for AI. It will also provide a basis for comparison of different approaches in the IT security domain beyond the project consortium.
Within the project, imbus is responsible for developing methods and tools for setting up realistic simulation environments that can serve as the basis for generating test and training data as well as for AI-based validation of the generated data. The know-how of imbus AG in the area of test automation, IT security as well as AI in general represents a great advantage within the research project.