Learning the learner: Using machine learning to track performance
summary
Anodot builds an automated monitoring layer that uses anomaly detection to track dozens of live models, claiming it instruments 30 algorithm types and watches on the order of 3 billion samples per day (tens of millions of time series updated every minute). The system combines online learners, clustering/correlation, and stacked autoencoders, links related metrics into behavioral topology, ingests semi‑supervised user feedback, and fires alerts (emails) that have caught both regressions from bad deploys and expected improvements such as seasonality fixes.
AI-written summary · about
chapters (7)
- 0:00The evolution of ML
- 2:01The ML process pipeline
- 4:29The monitoring gap
- 7:00Business incident detection
- 10:40Methods of anomaly detection
- 16:31Practical implementation
- 22:51Q&A and closing
related talks
28mHow Machine Learning Can Achieve Data Quality Monitoring At Scale
42mTesting Machine Learning Modules in Production
13mHuman-in-the-Loop MLOps: Boost Model Performance by 40%
25mMonitoring AI with AI
18mReliable Maintenance of Machine Learning Models
42mMachine Learning In Observability Systems: From Insights To Action
22mAI with Machine Learning for Observability in Netdata
01h 01mD1T2 - Applying Machine Learning to User Behavior Anomaly Analysis