Confident Approximate Policy Iteration for Efficient Local Planning in q \pi -realizable MDPs

Open in new window