In various experiments, our Bayesian SPNs often improve test likelihoods over greedy SPN learners.

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. This paper presents a framework for post model selection inference in the context of marginal screening, which is a computationally efficient way of variable selection in high dimensional linear regression problems. While the paper is focused on marginal screening, the approach is applicable to a broad range of problems. The paper is well written, very clear. I am not an expert in this area, but it seems like a significant problem and a statistically solid approach.

Neural Information Processing Systems http://nips.cc/

Time series constitute a challenging data type for machine learning algorithms, due to their highly variable lengths and sparse labeling in practice.

The Gumbel-Max trick is the basis of many relaxed gradient estimators. These estimators are easy to implement and low variance, but the goal of scaling them comprehensively to large combinatorial distributions is still outstanding.

The generated scene components overcome/impute occlusions (e.g. the purple