Safely Exploring Novel Actions in Recommender Systems via Deployment-Efficient Policy Learning

Open in new window