|First project meeting||17.10, 4pm||EMH
|Weekly project meeting||4pm||EMH
This semester, the CIT chair is offering a master project in the area of Artificial Intelligence for IT Operations. The goal is to develop an anomaly detection method on system data from distributed logs and metrics from Openstack.
The logging data represents interactions between data, files, or applications and is used to analyze specific trends or to record events/actions for a later forensic. The metric data reflects the current utilization and status of the infrastructure typically as cross-layer information regarding throughput, CPU, memory, disk, and latency. The most of the current AIOps platforms apply deep learning solely on monitoring data, as this data is simple to collect and interpret, but not sufficient for a holistic approach. The risk of fixing one detected anomaly and simultaneously causing additional problems due to the lack of holistic remediation plan, if frequently underestimated. Therefore, we aim at combining heterogeneous data sources and metrics for intelligent maintenance and troubleshooting to obtain a holistic view on a dynamic system and its current status and to select the globally effective self-healing activities.
This project aims at developing methods for robust anomaly detection from joint representation on multimodal system data (distributed logs and metrics). The algorithms to be investigated will allow intelligent data integration (fusion) of the different modalities. Finally, the goal is to utilize the joint representation in order to robustly and accurately detect anomalies that lead to component or system failures.
If you are interested in this project, please register via email (email@example.com ) and/or just join the kick-off meeting. In case you missed the meeting, please contact us anyway. We will probably have some places left!
Note: Please write "Master Project Anomaly Detection in Cloud" as your email subject.
Note that this project is an alternative to our "Master Project: Distributed Systems" (9LP). Hence, this course is part of the module "Master Project: Distributed Systems" (9LP) and can be used in the master studies Computer Science, Computer Engineering and Information Systems Management in the Study Area "Distributed Systems and Networks".