Ensemble learning is a method of combining multiple trained models to improve the model accuracy. We introduce the usage of such methods, specifically ensemble average inside Convolutional Neural Networks (CNNs) architectures. By Inner Average Ensemble (IEA) of multiple convolutional neural layers (CNLs) replacing the single CNLs inside the CNN architecture, the accuracy of the CNN increased. A visual and a similarity score analysis of the features generated from IEA explains why it boosts the model performance. Empirical results using different benchmarking datasets and well-known deep model architectures shows that IEA outperforms the ordinary CNL used in CNNs.

Graphical models have become a very popular tool for representing dependencies within a large set of variables and are key for representing causal structures. We provide results for uniform inference on high-dimensional graphical models with the number of target parameters being possible much larger than sample size. This is in particular important when certain features or structures of a causal model should be recovered. Our results highlight how in high-dimensional settings graphical models can be estimated and recovered with modern machine learning methods in complex data sets. We also demonstrate in simulation study that our procedure has good small sample properties.

Convolutional Neural Networks (CNNs) models have been recently introduced in the domain of top-$N$ session-based recommendations. An ordered collection of past items the user has interacted with in a session (or sequence) are embedded into a 2-dimensional latent matrix, and treated as an image. The convolution and pooling operations are then applied to the mapped item embeddings. In this paper, we first examine the typical session-based CNN recommender and show that both the generative model and network architecture are suboptimal when modeling long-range dependencies in the item sequence. To address the issues, we propose a simple, but very effective generative model that is capable of learning high-level representation from both short- and long-range dependencies. The network architecture of the proposed model is formed of a stack of holed convolutional layers, which can efficiently increase the receptive fields without relying on the pooling operation. Another contribution is the effective use of residual block structure in recommender systems, which can ease the optimization for much deeper networks. The proposed generative model attains state-of-the-art accuracy with less training time in the session-based recommendation task. It accordingly can be used as a powerful session-based recommendation baseline to beat in future, especially when there are long sequences of user feedback.

Multimodal learning has shown promising performance in content-based recommendation due to the auxiliary user and item information of multiple modalities such as text and images. However, the problem of incomplete and missing modality is rarely explored and most existing methods fail in learning a recommendation model with missing or corrupted modalities. In this paper, we propose LRMM, a novel framework that mitigates not only the problem of missing modalities but also more generally the cold-start problem of recommender systems. We propose modality dropout (m-drop) and a multimodal sequential autoencoder (m-auto) to learn multimodal representations for complementing and imputing missing modalities. Extensive experiments on real-world Amazon data show that LRMM achieves state-of-the-art performance on rating prediction tasks. More importantly, LRMM is more robust to previous methods in alleviating data-sparsity and the cold-start problem.

Empirical risk minimization (ERM), with proper loss function and regularization, is the common practice of supervised classification. In this paper, we study training arbitrary (from linear to deep) binary classifier from only unlabeled (U) data by ERM but not by clustering in the geometric space. A two-step ERM is considered: first an unbiased risk estimator is designed, and then the empirical training risk is minimized. This approach is advantageous in that we can also evaluate the empirical validation risk, which is indispensable for hyperparameter tuning when some validation data is split from U training data instead of labeled test data. We prove that designing such an estimator is impossible given a single set of U data, but it becomes possible given two sets of U data with different class priors. This answers a fundamental question in weakly-supervised learning, namely what the minimal supervision is for training any binary classifier from only U data. Since the proposed learning method is based on unbiased risk estimates, the asymptotic consistency of the learned classifier is certainly guaranteed. Experiments demonstrate that the proposed method could successfully train deep models like ResNet and outperform state-of-the-art methods for learning from two sets of U data.

Understanding nonlinear dynamical systems (NLDSs) is challenging in a variety of engineering and scientific fields. Dynamic mode decomposition (DMD), which is a numerical algorithm for the spectral analysis of Koopman operators, has been attracting attention as a way of obtaining global modal descriptions of NLDSs without requiring explicit prior knowledge. However, since existing DMD algorithms are in principle formulated based on the concatenation of scalar observables, it is not directly applicable to data with dependent structures among observables, which take, for example, the form of a sequence of graphs. In this paper, we formulate Koopman spectral analysis for NLDSs with structures among observables and propose an estimation algorithm for this problem. This method can extract and visualize the underlying low-dimensional global dynamics of NLDSs with structures among observables from data, which can be useful in understanding the underlying dynamics of such NLDSs. To this end, we first formulate the problem of estimating spectra of the Koopman operator defined in vector-valued reproducing kernel Hilbert spaces, and then develop an estimation procedure for this problem by reformulating tensor-based DMD. As a special case of our method, we propose the method named as Graph DMD, which is a numerical algorithm for Koopman spectral analysis of graph dynamical systems, using a sequence of adjacency matrices. We investigate the empirical performance of our method by using synthetic and real-world data.

Cyber-physical systems often consist of entities that interact with each other over time. Meanwhile, as part of the continued digitization of industrial processes, various sensor technologies are deployed that enable us to record time-varying attributes (a.k.a., time series) of such entities, thus producing correlated time series. To enable accurate forecasting on such correlated time series, this paper proposes two models that combine convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The first model employs a CNN on each individual time series, combines the convoluted features, and then applies an RNN on top of the convoluted features in the end to enable forecasting. The second model adds additional auto-encoders into the individual CNNs, making the second model a multi-task learning model, which provides accurate and robust forecasting. Experiments on two real-world correlated time series data set suggest that the proposed two models are effective and outperform baselines in most settings. This report extends the paper "Correlated Time Series Forecasting using Multi-Task Deep Neural Networks," to appear in ACM CIKM 2018, by providing additional experimental results.

Classification tasks usually assume that all possible classes are present during the training phase. This is restrictive if the algorithm is used over a long time and possibly encounters samples from unknown classes. The recently introduced extreme value machine, a classifier motivated by extreme value theory, addresses this problem and achieves competitive performance in specific cases. We show that this algorithm can fail when the geometries of known and unknown classes differ. To overcome this problem, we propose two new algorithms relying on approximations from extreme value theory. We show the effectiveness of our classifiers in simulations and on the LETTER and MNIST data sets.

Insurance companies must manage millions of claims per year. While most of these claims are non-fraudulent, fraud detection is core for insurance companies. The ultimate goal is a predictive model to single out the fraudulent claims and pay out the non-fraudulent ones immediately. Modern machine learning methods are well suited for this kind of problem. Health care claims often have a data structure that is hierarchical and of variable length. We propose one model based on piecewise feed forward neural networks (deep learning) and another model based on self-attention neural networks for the task of claim management. We show that the proposed methods outperform bag-of-words based models, hand designed features, and models based on convolutional neural networks, on a data set of two million health care claims. The proposed self-attention method performs the best.

Stochastic momentum methods have been widely adopted in training deep neural networks. However, their theoretical analysis of convergence of the training objective and the generalization error for prediction is still under-explored. This paper aims to bridge the gap between practice and theory by analyzing the stochastic gradient (SG) method, and the stochastic momentum methods including two famous variants, i.e., the stochastic heavy-ball (SHB) method and the stochastic variant of Nesterov's accelerated gradient (SNAG) method. We propose a framework that unifies the three variants. We then derive the convergence rates of the norm of gradient for the non-convex optimization problem, and analyze the generalization performance through the uniform stability approach. Particularly, the convergence analysis of the training objective exhibits that SHB and SNAG have no advantage over SG. However, the stability analysis shows that the momentum term can improve the stability of the learned model and hence improve the generalization performance. These theoretical insights verify the common wisdom and are also corroborated by our empirical analysis on deep learning.