You want to deploy a machine learning model in a GKE cluster. The model needs to be updated daily with minimal service disruption. What GKE feature should you use to accomplish this?

*Explanation:* Rolling updates are the ideal choice for deploying a machine learning model in a GKE cluster that requires daily updates with minimal service disruption. Rolling updates allow you to update the model while gradually replacing instances, minimizing disruption and ensuring a smooth daily update process. Other deployment strategies like blue-green deployments, canary deployments, and A/B testing are better suited for different scenarios, while horizontal pod autoscaling focuses on scaling, not updates.