Energy based models (EBMs) are appealing due to their generality and simplicity in likelihood modeling, but have been traditionally difficult to train. We present techniques to scale MCMC based EBM training, on continuous neural networks, and show its success on the high-dimensional data domains of ImageNet32x32, ImageNet128x128, CIFAR-10, and robotic hand trajectories, achieving significantly better samples than other likelihood models and on par with contemporary GAN approaches, while covering all modes of the data. We highlight unique capabilities of implicit generation, such as energy compositionality and corrupt image reconstruction and completion. Finally, we show that EBMs generalize well and are able to achieve state-of-the-art out-of-distribution classification, exhibit adversarially robust classification, coherent long term predicted trajectory roll-outs, and generate zero-shot compositions of models.

Clinical outcome prediction based on the Electronic Health Record (EHR) plays a crucial role in improving the quality of healthcare. Conventional deep sequential models fail to capture the rich temporal patterns encoded in the longand irregular clinical event sequences. We make the observation that clinical events at a long time scale exhibit strongtemporal patterns, while events within a short time period tend to be disordered co-occurrence. We thus propose differentiated mechanisms to model clinical events at different time scales. Our model learns hierarchical representationsof event sequences, to adaptively distinguish between short-range and long-range events, and accurately capture coretemporal dependencies. Experimental results on real clinical data show that our model greatly improves over previous state-of-the-art models, achieving AUC scores of 0.94 and 0.90 for predicting death and ICU admission respectively, Our model also successfully identifies important events for different clinical outcome prediction tasks

The fundamental role of neural networks (NNs) is readily apparent from their widespread use in machine learning in applications such as natural language processing [72], social network analysis [26], medical diagnosis [6, 35], vision systems [66], and robotic path planning [44]. The greatest success of these models lies in their flexibility, their ability to represent complex, nonlinear relationships in high-dimensional data sets, and the availability of frameworks that allow NNs to be implemented on rapidly evolving GPU platforms [40, 29]. The industrial appetite for deep learning has led to very rapid expansion of the subject in recent years, although, as pointed out by Dunson [19], at times the mathematical and theoretical understanding of these methods has been swept aside in the rush to advance the methodology. The potential impact on society of machine learning algorithms demands that their exposition and use be subject to the highest standards of clarity, ease of interpretation, and uncertainty quantification. Typical NN training seeks to optimize the parameters of the network (biases and weights) under the constraint that the training data set is well approximated [28, 23].

Standard stochastic optimization methods are brittle, sensitive to stepsize choices and other algorithmic parameters, and they exhibit instability outside of well-behaved families of objectives. To address these challenges, we investigate models for stochastic minimization and learning problems that exhibit better robustness to problem families and algorithmic parameters. With appropriately accurate models---which we call the aProx family---stochastic methods can be made stable, provably convergent and asymptotically optimal; even modeling that the objective is nonnegative is sufficient for this stability. We extend these results beyond convexity to weakly convex objectives, which include compositions of convex losses with smooth functions common in modern machine learning applications. We highlight the importance of robustness and accurate modeling with a careful experimental evaluation of convergence time and algorithm sensitivity.

We prove a convergence guarantee on the unadjusted Langevin algorithm for sampling assuming only that the target distribution $e^{-f}$ satisfies a log-Sobolev inequality and the Hessian of $f$ is bounded. In particular, $f$ is not required to be convex or have higher derivatives bounded.

Mini-batch sub-sampling is likely here to stay, due to growing data demands, memory-limited computational resources such as graphical processing units (GPUs), and the dynamics of on-line learning. Sampling a new mini-batch at every loss evaluation brings a number of benefits, but also one significant drawback: The loss function becomes discontinuous. These discontinuities are generally not problematic when using fixed learning rates or learning rate schedules typical of subgradient methods. However, they hinder attempts to directly minimize the loss function by solving for critical points, since function minimizers find spurious minima induced by discontinuities, while critical points may not even exist. Therefore, finding function minimizers and critical points in stochastic optimization is ineffective. As a result, attention has been given to reducing the effect of these discontinuities by means such as gradient averaging or adaptive and dynamic sampling. This paper offers an alternative paradigm: Recasting the optimization problem to rather find Non-Negative Associated Gradient Projection Points (NN-GPPs). In this paper, we demonstrate the NN-GPP interpretation of gradient information is more robust than critical points or minimizers, being less susceptible to sub-sampling induced variance and eliminating spurious function minimizers. We conduct a visual investigation, where we compare function value and gradient information for a variety of popular activation functions as applied to a simple neural network training problem. Based on the improved description offered by NN-GPPs over minimizers to identify true optima, in particular when using smooth activation functions with high curvature characteristics, we postulate that locating NN-GPPs can contribute significantly to automating neural network training.

In this paper we investigate two variants of association rules for preference data, Label Ranking Association Rules and Pairwise Association Rules. Label Ranking Association Rules (LRAR) are the equivalent of Class Association Rules (CAR) for the Label Ranking task. In CAR, the consequent is a single class, to which the example is expected to belong to. In LRAR, the consequent is a ranking of the labels. The generation of LRAR requires special support and confidence measures to assess the similarity of rankings. In this work, we carry out a sensitivity analysis of these similarity-based measures. We want to understand which datasets benefit more from such measures and which parameters have more influence in the accuracy of the model. Furthermore, we propose an alternative type of rules, the Pairwise Association Rules (PAR), which are defined as association rules with a set of pairwise preferences in the consequent. While PAR can be used both as descriptive and predictive models, they are essentially descriptive models. Experimental results show the potential of both approaches.

Deep neural networks have outperformed existing machine learning models in various molecular applications. In practical applications, it is still difficult to make confident decisions because of the uncertainty in predictions arisen from insufficient quality and quantity of training data. Here, we show that Bayesian neural networks are useful to quantify the uncertainty of molecular property prediction with three numerical experiments. In particular, it enables us to decompose the predictive variance into the model- and data-driven uncertainties, which helps to elucidate the source of errors. In the logP predictions, we show that data noise affected the data-driven uncertainties more significantly than the model-driven ones. Based on this analysis, we were able to find unexpected errors in the Harvard Clean Energy Project dataset. Lastly, we show that the confidence of prediction is closely related to the predictive uncertainty by performing on bio-activity and toxicity classification problems.

The rise of non-linear and interactive media such as video games has increased the need for automatic movement animation generation. In this survey, we review and analyze different aspects of building automatic movement generation systems using machine learning techniques and motion capture data. We cover topics such as high-level movement characterization, training data, features representation, machine learning models, and evaluation methods. We conclude by presenting a discussion of the reviewed literature and outlining the research gaps and remaining challenges for future work.

A classic setting of the stochastic K-armed bandit problem is considered in this note. In this problem it has been known that KL-UCB policy achieves the asymptotically optimal regret bound and KL-UCB policy empirically performs better than the KL-UCB policy although the regret bound for the original form of the KL-UCB policy has been unknown. This note demonstrates that a simple proof of the asymptotic optimality of the KL-UCB policy can be given by the same technique as those used for analyses of other known policies. In the problem of the stochastic bandit problems, it is known that there exists a (problem-dependent) regret lower bound [1][2]. It can be achieved by, for example, the DMED policy [3] for the model of nonparametric distributions over [0, 1]. One of the conference version [6] of [5] also proposed KL-UCB policy, which empirically performs better than KL-UCB but does not have a theoretical guarantee.