Efficient pooling of predictions via kernel embeddings

Open in new window