This paper addresses a potentially important problem by giving an algorithm that can solve large constrained POMDPs with online methods. A constrained POMDP, which augments a traditional POMDP with multi-attribute cost constraints, is an important extension that can help model a wider range of real-world phenomena than a POMDP can. Having such an algorithm for solving large CPOMDPs is a very valuable contribution. The authors provide, in this paper, a derivation of an unconstrained objective to be solved (resulting from taking the dual of the CPOMDP's linear program), backed by theoretical justification, and an adaptation of the online search algorithm, POMCP, that incorporates cost constraints by approximately optimizing the objective. The paper is extremely well-written, free of typos, and clear in its presentation.

Main ideas The paper develops the relation between solving an MDP and performing inference in a Bayesian network. The direction, however, is novel as far as I can tell: using MDP algorithms to solve an inference problem. The first part shows that an existing MDP algorithm (ARollout) is in fact performing a BP iteration over the DBN that represents the MDP. In the second part, a different MDP algorithm (SOGBOFA) is used to solve a particular inference problem of choosing a subset of values with the maximal marginals (MMAP). The resulting SOGBOFA-based solver often loses to the state-of-the-art, but for harder cases it can outperform the state of the art.

This paper presents Deep Poisson-Gamma Dynamical System (DPGDS) for modeling temporal multivariate count data. It is based on previously developed Gamma-Belief networks, extended to the dynamical scenarios by adding transitions of latent units in consecutive times. The paper is well written, and connections to previous papers are explained clearly. While the temporal structure is based on transition of latent units in a Markov manner, the authors' claim about better capturing long-range temporal changes should be justified more clearly. In line 30, "separated" should be replaced by "separately".

The manuscript "Learning Concave Conditional Likelihood Models for Improved Analysis of Tandem Mass Spectra" extends a dynamic Bayesian network approach called DIDEA by introducing a new class of emission distributions. The conditional log-likelihood of those functions remains concave leading to an efficient global optimization method for parameter estimation. This is in stark contrast to the previous variant, for which the best parameter had to be found by grid search. In comparison to other state-of-the-art methods, the new approach outperforms the other methods, while being faster at the same time. Quality Overall the quality of the manuscript is good.

This paper is very interesting. One previous assumption in TD learning is that reward are close with states in proximity of the state space, which has been pointed out by many papers is not realistic and have problems for spatial value function regularization. Instead, this paper make the assumption that rewards are close for states. Overall this paper has a very good motivation, and the literature review shows that the author is knowledgable of this field. This paper could open a novel area of temporal regularization that received inadequate attention before.

I preface this by saying that I have reviewed this paper once for NIPS 2016, and re-read it. It seems the paper has no essential changes, so my opinion is largely the same. The paper considers the problem of learning the parameters of a Bayes net with known structure, given samples from it with potentially adversarial noise. The main goal is to get bounds on the samples that are independent of dimension. The main requirements on the Bayes net parameters are reasonable: the probability of any configuration of the parents is reasonable and the conditional probabilities on any edge are bounded away from 0 and 1.

This is an excellent theoretical contribution. The analysis is quite heavy and has many subtleties. I do not have enough time to read the appended proofs; also, the subject of the paper is not in my area of research. The comments below are based on the impression I got after reading carefully the first 8 pages of the paper and glancing through the rest in the supplementary file. Summary: This paper is about reinforcement learning in weakly-communicating MDP under the average-reward criterion.

This paper puts forward the idea that we should in certain cases regularize the generative model in VAEs in order to improve generalization properties. Since VAEs perform maximum likelihood estimation, they can in principle exhibit the same overfitting problems as any other maximum likelihood model. This paper argues that we can regularize the generative model by increasing the smoothness of the inference model. The authors consider the Denoising VAE (DVAE) as a means of achieving such regularization. In the special case where the encoder is an exponential family, they show that the optimum natural parameters for any input data can be expressed as a weighted average over the optimum parameters for the data in the training set.

POST REBUTTAL: I do think that the edit to the proof suggested by the authors could work, but would lead to some exorbitant constant C4, a subject not addressed by the authors. Still, I have increased my score from "clear reject" to "accept" in the light of the fact that I am now happy with the validity of the proofs.

This work expands on the algorithm RAI by Yehezkel and Lerner for constraint-based structure learning of Bayesian networks. Each node of the tree splits the variables into subsets (one descendant and K ancestral subsets) by using conditional independence (CI) tests of order n. The submission proposes the B-RAI algorithm that leverages bootstrap to allow the algorithm to output a set of highly likely CPDAG rather than the MAP one. Bootstrap is not naively leveraged. Instead, it is integrated in the recursive call of the algorithm.