From the course: Cloud-Based AI Solution Design Patterns
Join today to access over 24,700 courses taught by industry experts.
AI-model drift detection
From the course: Cloud-Based AI Solution Design Patterns
AI-model drift detection
- The AI model drift detection pattern introduces a separate set of cloud-native, model-specific monitors that focus solely on detecting concept drift and data drift. Let's quickly recap these terms before we begin. Concept drift happens when the manner in which a model generates output differs from the input data it is receiving. Data drift occurs when the actual input data is different than what the model was trained to handle. For example, concept drift might happen when the customer data being received includes different statistics than what the model was trained to process in order to predict customer churn, whereas data drift would be happening when the model begins receiving weather forecast data instead of customer profile data altogether. So let's begin with the concept drift monitor, which is tasked with continually checking the input data received by an AI system and then comparing it with the data the model was originally trained with. If the incoming data starts looking…
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.
