1.Techniques 1.1 Anomaly detection and machine learning are promising approaches for monitoring. 1.2 Dedicated anomaly detection methods perform better. 1.3 Learning methods that combine data-driven and knowledge-based approaches have the most potential. 2.Approach 2.1 Machine learning (or data science) is an incremental, joint effort. Close interaction between Al experts and application specialists. 2.2 Feedback is crucial (labels, logs, …). 2.3 You need highly motivated collaborators (thank you Thomas, Elien, Dragos, David and all others).
展开▼