From the course: AI for Telecom: Network Optimization and Security in 5G/Edge Systems
Join today to access over 24,600 courses taught by industry experts.
Operations phase in AI lifecycle
From the course: AI for Telecom: Network Optimization and Security in 5G/Edge Systems
Operations phase in AI lifecycle
- [Instructor] AI's a real test begins after deployment when it faces live networks and real users. The operation phase is where models go from smart to street smart. In this session, we'll go through model serving and deployment, orchestration, as well as we'll show you how you can effectively monitor the KPIs. Let's begin. In the model serving, it's like taking a trained AI model and making it available in real time so that telecom systems can use its prediction. There are certain stages to it. It includes containerization where you have to package the model. Then we have model servers where you can use tools like TensorFlow Serving, TorchServe or ONAP machine learning framework to expose the predictions via the APIs. And you can also use integration where you can plug the model into the real time data pipelines. Then it is scalability and resilience, where if you're running on Kubernetes, the load can be effectively handled. For monitoring, track latency, accuracy, and drift in the…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.