Questions on whether governments have acted promptly enough, and whether lockdown measures can be lifted soon, have since been central in public discourse. Data-driven models that predict COVID-19 fatalities under different lockdown policy scenarios are essential for addressing these questions and informing governments on future policy directions.

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. The goal of this work is to automatically discover latent domains in a training set, which is subsequently used in a domain adaptation framework to yield improved classification performance on a test set. The paper defines a function that measures the difference between two feature vectors over a specified kernel. The goal is to partition the data points into domains such that the function is maximized over the set of points across each pair of domains. The problem is formulated as an integer programming problem with two constraints: each point is assigned to exactly one domain and the distribution over class labels in each domain must match the input distribution over the entire point set.

We consider the problem of identifying multiway block structure from a large noisy tensor. Such problems arise frequently in applications such as genomics, recommendation system, topic modeling, and sensor network localization.

We thank the reviewers for the helpful comments and feedback. Our responses are detailed below. We will make the suggested edits for clarity. The improved interpretability with little loss of accuracy makes the sparse TBM appealing in applications. We agree with reviewer that MSE is not the best metric for clustering.

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. 'Learning Gaussian Graphical Models with Observed or Latent FVSs' addresses the problem of learning (i.e. The motivation is that exact inference under these models can be done quickly, and so in the case where one needs near-linear inference (which is prohibited in general for sparse GGMs) it is desirable to have this form. The results address three cases: (4.1.1) In (4.1.2) they make the observation that one can exhaustively run the previous algorithm for all k-sets selecting the one that maximizes the likelihood and then provide a greedy algorithm.

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

This objective is amenable to minibatching. The variational posterior tracks the true posterior during gradient updates.