Language system dynamics is studied at several distinct time scales (Lewis, 2014).

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. This paper considers transfer learning in a multi-armed bandit setting. The model considered has a sequence of episodes, and in each episode, the vector of distributions (one for each arm) is drawn iid from a discrete distribution. In this setting, it is possible to exploit history to learn what this discrete distribution is, and to use this information to reduce regret in each episode. An algorithm is proposed that does this, and cumulative regret bounds are shown for this algorithm.

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. The paper proposes a method for relating images and sentences by optimizing over the mapping of sentence fragments and image regions. The method uses existing word vector and image region representations. Experiments show that this method is better able to rank human-generated image captions. Quality The paper is high quality, and well-supported by empirical results.

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. Message Passing Inference for Large Scale Graphical Models with High Order Potentials The paper follows up on recent work on parallelizing message passing algorithms. The main contribution is dealing with higher order potentials. The main insight is that if large potentials are unary the message they send out is constant. By adding auxiliary factors this can be exploited (Fig 2 gives an example).

However, knowing which data to present during different stages of training is a challenging problem.

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. The authors combine two recent advances in Gaussian processes, spectral mixture kernels [5], and scalable Gaussian processes for data in grids [14, 22 see below], in order to tackle applications with high amount of data points, like texture extrapolation, inpainting, and video extrapolation. The paper includes a thorough evaluation of the framework proposed, and comparisons against sparse GP methods, with general purpose covariance functions, and spectral mixture kernels. Quality The paper is technically sound. The framework proposed by the authors achieves outstanding results in the different applications studied in the paper.

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. Paper Summary: This paper treats a general multi-armed bandit problem in which the mean reward of each arm depends on a common unknown parameter. The authors consider a simple modification of the UCB1 algorithm. They show, unsurprisingly, that the algorithm satisfies a regret bound like that of UCB1. The main improvement of this paper is to show when the optimal arm can be identified perfectly by samples of the optimal arm, algorithm's regret is bounded by a constant independent of the time horizon.

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. Line 33: I don't think it is accurate to attribute the recent success of supervised neural nets on various applications to BP and dropout. Firstly, learning nets with gradient descent has been around a long time, and the key to its recent success has mostly been fast computers/GPUs, a wealth of labelled data, advances in understanding of how to make SGD work well (e.g. Techniques like dropout have also been useful in reducing overfitting, but are hardly the key missing ingredient to make these systems work well. Line 37: The claim that the lacklustre the results associated with unsupervised generative approaches is owed purely to their intractability issues is a strong and problematic one.

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. The authors present a variational approach L-FIELD to general log-submodular and supermodular distributions. Theoretical contributions include deriving upper and lower bounds on the log-partition function and fully factorized approximate posteriors. The quality of the approximation is tested with respect to the curvature of the function. Empirical results are presented on GMM cuts and MRFs, decomposable functions and facility location modeling.

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. The paper considers the setting of a sensor network (or agents) in a noisy environment that are able to communicate locally. The authors prove that theoretical bounds on the number of active queries can be achieved through simple best response dynamics. The paper is very well-written, technically correct, and the synthetic experiment makes the results clear. The theoretical results are novel and I think that the paper deserves to be published to be published at NIPS.