Convolutional Neural Networks (CNNs) are known to rely more on local texture rather than global shape when making decisions. Recent work also indicates a close relationship between CNN's texture-bias and its robustness against distribution shift, adversarial perturbation, random corruption, etc. In this work, we attempt at improving various kinds of robustness universally by alleviating CNN's texture bias. With inspiration from the human visual system, we propose a light-weight model-agnostic method, namely Informative Dropout (InfoDrop), to improve interpretability and reduce texture bias. Specifically, we discriminate texture from shape based on local self-information in an image, and adopt a Dropout-like algorithm to decorrelate the model output from the local texture. Through extensive experiments, we observe enhanced robustness under various scenarios (domain generalization, few-shot classification, image corruption, and adversarial perturbation). To the best of our knowledge, this work is one of the earliest attempts to improve different kinds of robustness in a unified model, shedding new light on the relationship between shape-bias and robustness, also on new approaches to trustworthy machine learning algorithms. Code is available at https://github.com/bfshi/InfoDrop.

EagerPy is a Python framework that lets you write code that automatically works natively with PyTorch, TensorFlow, JAX, and NumPy. Library developers no longer need to choose between supporting just one of these frameworks or reimplementing the library for each framework and dealing with code duplication. Users of such libraries can more easily switch frameworks without being locked in by a specific 3rd party library. Beyond multi-framework support, EagerPy also brings comprehensive type annotations and consistent support for method chaining to any framework. The latest documentation is available online at https://eagerpy.jonasrauber.de and the code can be found on GitHub at https://github.com/jonasrauber/eagerpy.

Deep learning models have achieved expert-level performance in healthcare with an exclusive focus on training accurate models. However, in many clinical environments such as intensive care unit (ICU), real-time model serving is equally if not more important than accuracy, because in ICU patient care is simultaneously more urgent and more expensive. Clinical decisions and their timeliness, therefore, directly affect both the patient outcome and the cost of care. To make timely decisions, we argue the underlying serving system must be latency-aware. To compound the challenge, health analytic applications often require a combination of models instead of a single model, to better specialize individual models for different targets, multi-modal data, different prediction windows, and potentially personalized predictions. To address these challenges, we propose HOLMES-an online model ensemble serving framework for healthcare applications. HOLMES dynamically identifies the best performing set of models to ensemble for highest accuracy, while also satisfying sub-second latency constraints on end-to-end prediction. We demonstrate that HOLMES is able to navigate the accuracy/latency tradeoff efficiently, compose the ensemble, and serve the model ensemble pipeline, scaling to simultaneously streaming data from 100 patients, each producing waveform data at 250~Hz. HOLMES outperforms the conventional offline batch-processed inference for the same clinical task in terms of accuracy and latency (by order of magnitude). HOLMES is tested on risk prediction task on pediatric cardio ICU data with above 95% prediction accuracy and sub-second latency on 64-bed simulation.

An essential task in predictive maintenance is the prediction of the Remaining Useful Life (RUL) through the analysis of multivariate time series. Using the sliding window method, Convolutional Neural Network (CNN) and conventional Recurrent Neural Network (RNN) approaches have produced impressive results on this matter, due to their ability to learn optimized features. However, sequence information is only partially modeled by CNN approaches. Due to the flatten mechanism in conventional RNNs, like Long Short Term Memories (LSTM), the temporal information within the window is not fully preserved. To exploit the multi-level temporal information, many approaches are proposed which combine CNN and RNN models. In this work, we propose a new LSTM variant called embedded convolutional LSTM (ECLSTM). In ECLSTM a group of different 1D convolutions is embedded into the LSTM structure. Through this, the temporal information is preserved between and within windows. Since the hyper-parameters of models require careful tuning, we also propose an automated prediction framework based on the Bayesian optimization with hyperband optimizer, which allows for efficient optimization of the network architecture. Finally, we show the superiority of our proposed ECLSTM approach over the state-of-the-art approaches on several widely used benchmark data sets for RUL Estimation.

Machine learning can provide deep insights into data, allowing machines to make high-quality predictions and having been widely used in real-world applications, such as text mining, visual classification, and recommender systems. However, most sophisticated machine learning approaches suffer from huge time costs when operating on large-scale data. This issue calls for the need of {Large-scale Machine Learning} (LML), which aims to learn patterns from big data with comparable performance efficiently. In this paper, we offer a systematic survey on existing LML methods to provide a blueprint for the future developments of this area. We first divide these LML methods according to the ways of improving the scalability: 1) model simplification on computational complexities, 2) optimization approximation on computational efficiency, and 3) computation parallelism on computational capabilities. Then we categorize the methods in each perspective according to their targeted scenarios and introduce representative methods in line with intrinsic strategies. Lastly, we analyze their limitations and discuss potential directions as well as open issues that are promising to address in the future.

In this paper, we study the problem of learning continuous vector representations of knowledge graphs for predicting missing links. We present a new approach called ConEx, which infers missing links by leveraging the composition of a 2D convolution with a Hermitian inner product of complex-valued embedding vectors. We evaluate ConEx against state-of-the-art approaches on the WN18RR, FB15K-237, KINSHIP and UMLS benchmark datasets. Our experimental results show that ConEx achieves a performance superior to that of state-of-the-art approaches such as RotatE, QuatE and TuckER on the link prediction task on all datasets while requiring at least 8 times fewer parameters. We ensure the reproducibility of our results by providing an open-source implementation which includes the training, evaluation scripts along with pre-trained models at https://github.com/conex-kge/ConEx.

Deep neural networks have become popular for many machine learning applications, since they provide simple strategies for end-to-end learning of complex representations. However, success can be highly sensitive to network architectures, which places a great demand on manual engineering of architectures and hyper-parameter tuning. A typical human-designed convolutional neural architecture is composed of two types of computational modules: i) a normal layer, such as a stride-1 convolution or an identity mapping, which maintains the spatial dimension of incoming feature maps; ii) a resizing layer, such as max/average pooling, bilinear sampling, or stride-2 convolution, which reshapes the incoming feature map into a different spatial dimension. We hereby define the shape of a neural network as the composition of the feature dimensions in all network layers, and the architecture as the overall structure formed by stacking multiple normal and resizing layers.

Forecasting groups of time series is of increasing practical importance, e.g. forecasting the demand for multiple products offered by a retailer or server loads within a data center. The local approach to this problem considers each time series separately and fits a function or model to each series. The global approach fits a single function to all series. For groups of similar time series, global methods outperform the more established local methods. However, recent results show good performance of global models even in heterogeneous datasets. This suggests a more general applicability of global methods, potentially leading to more accurate tools and new scenarios to study. Formalizing the setting of forecasting a set of time series with local and global methods, we provide the following contributions: 1) Global methods are not more restrictive than local methods, both can produce the same forecasts without any assumptions about similarity of the series. Global models can succeed in a wider range of problems than previously thought. 2) Basic generalization bounds for local and global algorithms. The complexity of local methods grows with the size of the set while it remains constant for global methods. In large datasets, a global algorithm can afford to be quite complex and still benefit from better generalization. These bounds serve to clarify and support recent experimental results in the field, and guide the design of new algorithms. For the class of autoregressive models, this implies that global models can have much larger memory than local methods. 3) In an extensive empirical study, purposely naive algorithms derived from these principles, such as global linear models or deep networks result in superior accuracy. In particular, global linear models can provide competitive accuracy with two orders of magnitude fewer parameters than local methods.

Contrastive learning and supervised learning have both seen significant progress and success. However, thus far they have largely been treated as two separate objectives, brought together only by having a shared neural network. In this paper we show that through the perspective of hybrid discriminative-generative training of energy-based models we can make a direct connection between contrastive learning and supervised learning. Beyond presenting this unified view, we show our specific choice of approximation of the energy-based loss outperforms the existing practice in terms of classification accuracy of WideResNet on CIFAR-10 and CIFAR-100. It also leads to improved performance on robustness, out-of-distribution detection, and calibration.

We study differentially private (DP) algorithms for stochastic non-convex optimization. In this problem, the goal is to minimize the population loss over a $p$-dimensional space given $n$ i.i.d. samples drawn from a distribution. We improve upon the population gradient bound of ${\sqrt{p}}/{\sqrt{n}}$ from prior work and obtain a sharper rate of $\sqrt[4]{p}/\sqrt{n}$. We obtain this rate by providing the first analyses on a collection of private gradient-based methods, including adaptive algorithms DP RMSProp and DP Adam. Our proof technique leverages the connection between differential privacy and adaptive data analysis to bound gradient estimation error at every iterate, which circumvents the worse generalization bound from the standard uniform convergence argument. Finally, we evaluate the proposed algorithms on two popular deep learning tasks and demonstrate the empirical advantages of DP adaptive gradient methods over standard DP SGD.