Modellbereitstellung (EN)

Modellbereitstellung

The model deployment is the process of implementing trained machine learning models in production environments. This process includes various steps such as model serialization, API creation, and integration into existing systems. An efficient and automated model deployment is a core goal of MLOps to reduce the time-to-market for ML applications.

Architecture

flowchart TD     A[Trained ML Model] --> B[Model Serialization]     B --> C[Containerization]     C --> D[API Creation]     D --> E[CI/CD Pipeline]     E --> F[Staging Environment]     F --> G[Production Environment]     G --> H[Monitoring & Logging]     H --> I[Feedback Loop]     I --> A 

In Context

• Typically used together with MLOps, Continuous Integration/Continuous Deployment (CI/CD) and Model Monitoring
• Related to: Model Versioning, A/B Testing, Canary Deployment, Blue-Green Deployment
• Example use case: An image recognition model is deployed as a REST API called by a mobile application