Trajectory optimization with learned dynamics models can often suffer from erroneous predictions of out-of-distribution trajectories. We propose to regularize trajectory optimization by means of a denoising autoencoder that is trained on the same trajectories as the dynamics model. We visually demonstrate the effectiveness of the regularization in gradient-based trajectory optimization for open-loop control of an industrial process. We compare with recent model-based reinforcement learning algorithms on a set of popular motor control tasks to demonstrate that the denoising regularization enables state-of-the-art sample-efficiency. We demonstrate the efficacy of the proposed method in regularizing both gradient-based and gradient-free trajectory optimization.

Patients with Type I Diabetes (T1D) must take insulin injections to prevent the serious long term effects of hyperglycemia - high blood glucose (BG). Patients must also be careful not to inject too much insulin because this could induce hypoglycemia (low BG), which can potentially be fatal. Patients therefore follow a "regimen" that determines how much insulin to inject at certain times. Current methods for managing this disease require adjusting the patient's regimen over time based on the disease's behavior (recorded in the patient's diabetes diary). If we can accurately predict a patient's future BG values from his/her current features (e.g., predicting today's lunch BG value given today's diabetes diary entry for breakfast, including insulin injections, and perhaps earlier entries), then it is relatively easy to produce an effective regimen. This study explores the challenges of BG modeling by applying several machine learning algorithms and various data preprocessing variations (corresponding to 312 [learner, preprocessed-dataset] combinations), to a new T1D dataset containing 29 601 entries from 47 different patients. Our most accurate predictor is a weighted ensemble of two Gaussian Process Regression models, which achieved an errL1 loss of 2.70 mmol/L (48.65 mg/dl). This was an unexpectedly poor result given that one can obtain an errL1 of 2.91 mmol/L (52.43 mg/dl) using the naive approach of simply predicting the patient's average BG. For each of data-variant/model combination we report several evaluation metrics, including glucose-specific metrics, and find similarly disappointing results (the best model was only incrementally better than the simplest measure). These results suggest that the diabetes diary data that is typically collected may not be sufficient to produce accurate BG prediction models; additional data may be necessary to build accurate BG prediction models.

Humans tend to learn complex abstract concepts faster if examples are presented in a structured manner. For instance, when learning how to play a board game, usually one of the first concepts learned is how the game ends, i.e. the actions that lead to a terminal state (win, lose or draw). The advantage of learning end-games first is that once the actions which lead to a terminal state are understood, it becomes possible to incrementally learn the consequences of actions that are further away from a terminal state - we call this an end-game-first curriculum. Currently the state-of-the-art machine learning player for general board games, AlphaZero by Google DeepMind, does not employ a structured training curriculum; instead learning from the entire game at all times. By employing an end-game-first training curriculum to train an AlphaZero inspired player, we empirically show that the rate of learning of an artificial player can be improved during the early stages of training when compared to a player not using a training curriculum.

For sampling from a log-concave density, we study implicit integrators resulting from $\theta$-method discretization of the overdamped Langevin diffusion stochastic differential equation. Theoretical and algorithmic properties of the resulting sampling methods for $ \theta \in [0,1] $ and a range of step sizes are established. Our results generalize and extend prior works in several directions. In particular, for $\theta\ge1/2$, we prove geometric ergodicity and stability of the resulting methods for all step sizes. We show that obtaining subsequent samples amounts to solving a strongly-convex optimization problem, which is readily achievable using one of numerous existing methods. Numerical examples supporting our theoretical analysis are also presented.

Enforcing distributions of latent variables in neural networks is an active subject. It is vital in all kinds of generative models, where we want to be able to interpolate between points in the latent space, or sample from it. Modern generative AutoEncoders (AE) like WAE, SWAE, CWAE add a regularizer to the standard (deterministic) AE, which allows to enforce Gaussian distribution in the latent space. Enforcing different distributions, especially topologically nontrivial, might bring some new interesting possibilities, but this subject seems unexplored so far. This article proposes a new approach to enforce uniform distribution on d-dimensional torus. We introduce a circular spring loss, which enforces minibatch points to be equally spaced and satisfy cyclic boundary conditions. As example of application we propose multiple-path morphing. Minimal distance geodesic between two points in uniform distribution on latent space of angles becomes a line, however, torus topology allows us to choose such lines in alternative ways, going through different edges of $[-\pi,\pi]^d$. Further applications to explore can be for example trying to learn real-life topologically nontrivial spaces of features, like rotations to automatically recognize 2D rotation of an object in picture by training on relative angles, or even 3D rotations by additionally using spherical features - this way morphing should be close to object rotation.

In this paper we establish rigorous benchmarks for image classifier robustness. MAGENET-C,standardizes and expands the corruption robustness topic, while showing which classifiers are preferable in safety-critical applications. MAGENET-Pwhich enables researchers to benchmark a classifier's robustness to common perturbations. Unlike recent robustness research, this benchmark evaluates performance on common corruptions and perturbations not worst-case adversarial perturbations. We find that there are negligible changes in relative corruption robustness from AlexNet classifiers to ResNet classifiers. Afterward we discover ways to enhance corruption and perturbation robustness. We even find that a bypassed adversarial defense provides substantial common perturbation robustness. Together our benchmarks may aid future work toward networks that robustly generalize. The human vision system is robust in ways that existing computer vision systems are not (Recht et al., 2018; Azulay & Weiss, 2018). Unlike current deep learning classifiers (Krizhevsky et al., 2012; He et al., 2015; Xie et al., 2016), the human vision system is not fooled by small changes in query images. Humans are also not confused by many forms of corruption such as snow, blur, pixelation, and novel combinations of these. Humans can even deal with abstract changes in structure and style. Achieving these kinds of robustness is an important goal for computer vision and machine learning.

Optical scatterometry is a method to measure the size and shape of periodic micro- or nanostructures on surfaces. For this purpose the geometry parameters of the structures are obtained by reproducing experimental measurement results through numerical simulations. We compare the performance of Bayesian optimization to different local minimization algorithms for this numerical optimization problem. Bayesian optimization uses Gaussian-process regression to find promising parameter values. We examine how pre-computed simulation results can be used to train the Gaussian process and to accelerate the optimization.

In this work, we utilize Machine Learning for early recognition of patients at high risk of acute respiratory distress syndrome (ARDS), which is critical for successful prevention strategies for this devastating syndrome. The difficulty in early ARDS recognition stems from its complex and heterogenous nature. In this study, we integrate knowledge of the heterogeneity of ARDS patients into predictive model building. Using MIMIC-III data, we first apply latent class analysis (LCA) to identify homogeneous sub-groups in the ARDS population, and then build predictive models on the partitioned data. The results indicate that significantly improved performances of prediction can be obtained for two of the three identified sub-phenotypes of ARDS. Experiments suggests that identifying sub-phenotypes is beneficial for building predictive model for ARDS.

Automatic speech emotion recognition provides computers with critical context to enable user understanding. While methods trained and tested within the same dataset have been shown successful, they often fail when applied to unseen datasets. To address this, recent work has focused on adversarial methods to find more generalized representations of emotional speech. However, many of these methods have issues converging, and only involve datasets collected in laboratory conditions. In this paper, we introduce Adversarial Discriminative Domain Generalization (ADDoG), which follows an easier to train "meet in the middle" approach. The model iteratively moves representations learned for each dataset closer to one another, improving cross-dataset generalization. We also introduce Multiclass ADDoG, or MADDoG, which is able to extend the proposed method to more than two datasets, simultaneously. Our results show consistent convergence for the introduced methods, with significantly improved results when not using labels from the target dataset. We also show how, in most cases, ADDoG and MADDoG can be used to improve upon baseline state-of-the-art methods when target dataset labels are added and in-the-wild data are considered. Even though our experiments focus on cross-corpus speech emotion, these methods could be used to remove unwanted factors of variation in other settings.

In information retrieval (IR) and related tasks, term weighting approaches typically consider the frequency of the term in the document and in the collection in order to compute a score reflecting the importance of the term for the document. In tasks characterized by the presence of training data (such as text classification) it seems logical that the term weighting function should take into account the distribution (as estimated from training data) of the term across the classes of interest. Although `supervised term weighting' approaches that use this intuition have been described before, they have failed to show consistent improvements. In this article we analyse the possible reasons for this failure, and call consolidated assumptions into question. Following this criticism we propose a novel supervised term weighting approach that, instead of relying on any predefined formula, learns a term weighting function optimised on the training set of interest; we dub this approach \emph{Learning to Weight} (LTW). The experiments that we run on several well-known benchmarks, and using different learning methods, show that our method outperforms previous term weighting approaches in text classification.