Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
After a discussion among reviewers, this submission has been considered a solid theoretical contribution to the specific problem of learning polytree networks (with latent variables). It is important to understand what can be learned. The work also proposes algorithms to recover the network under some conditions. On the downside, there is no good motivations and no evaluation of any kind. The latter is hard to be resolved quickly, but an effort could be put in improving the motivations and potential impact.