Lightning Talk: Outsourcing Telco Understanding AI
Lightningat Cloud Native Telco Day North America 2023 · on Monday, 2023-11-06· at 14:45· W193ab, McCormick Place·
summary
Cloud-native 5G infrastructure generates petabytes per day across 18,000 edge nodes, so machines learn patterns rather than humans. An autoencoder on EKS metrics and application logs flags atypical residuals, and a reinforcement-learning agent—First Open Network Pattern Reactor (FONPR)—is trained to take scale-up/scale-down actions guided by a v1 reward balancing service continuity and cost. Code and research are open source on GitHub and Medium; the system is already detecting patterns and proving automated reactions in lab and pilot deployments.
AI-written summary · about
chapters (9)
- 0:00Self-perfecting networks intro
- 1:06Cloud-native 5G infrastructure
- 2:10Open APIs and security
- 3:34Enterprise flexible control
- 4:37Scaling with AI and RL
- 6:06Autoencoders for pattern detection
- 7:52FPER for autonomous reactions
- 9:39AI-driven network future
- 10:43Open source contributions
related talks
23mKeynote: Network Intelligence and Open Source: Practice and Thoughts
20mCloud-Native AI: Architectural Patterns for Enterprise Success
23mThe State of GenAI & ML in the Cloud Native Ecosystem
34mLFN AI Taskforce: Platform-Driven Fede...
21mFrom Reactive to Proactive: How AI is Redefining Resilience in Modern Networks
33mEdge-Native: The new paradigm for operating and developing Edge Apps
54mMachine Learning for Developers
45mAIoT Ops - Using Kubernetes and ML Ops to Build Edge ML Applications