This paper demonstrates how to apply machine learning algorithms to distinguish good stocks from the bad stocks. To this end, we construct 244 technical and fundamental features to characterize each stock, and label stocks according to their ranking with respect to the return-to-volatility ratio. Algorithms ranging from traditional statistical learning methods to recently popular deep learning method, e.g. Logistic Regression (LR), Random Forest (RF), Deep Neural Network (DNN), and the Stacking, are trained to solve the classification task. Genetic Algorithm (GA) is also used to implement feature selection. The effectiveness of the stock selection strategy is validated in Chinese stock market in both statistical and practical aspects, showing that: 1) Stacking outperforms other models reaching an AUC score of 0.972; 2) Genetic Algorithm picks a subset of 114 features and the prediction performances of all models remain almost unchanged after the selection procedure, which suggests some features are indeed redundant; 3) LR and DNN are radical models; RF is risk-neutral model; Stacking is somewhere between DNN and RF. 4) The portfolios constructed by our models outperform market average in back tests.

Graphs are widely used as a natural framework that captures interactions between individual elements represented as nodes in a graph. In medical applications, specifically, nodes can represent individuals within a potentially large population (patients or healthy controls) accompanied by a set of features, while the graph edges incorporate associations between subjects in an intuitive manner. This representation allows to incorporate the wealth of imaging and non-imaging information as well as individual subject features simultaneously in disease classification tasks. Previous graph-based approaches for supervised or unsupervised learning in the context of disease prediction solely focus on pairwise similarities between subjects, disregarding individual characteristics and features, or rather rely on subject-specific imaging feature vectors and fail to model interactions between them. In this paper, we present a thorough evaluation of a generic framework that leverages both imaging and non-imaging information and can be used for brain analysis in large populations. This framework exploits Graph Convolutional Networks (GCNs) and involves representing populations as a sparse graph, where its nodes are associated with imaging-based feature vectors, while phenotypic information is integrated as edge weights. The extensive evaluation explores the effect of each individual component of this framework on disease prediction performance and further compares it to different baselines. The framework performance is tested on two large datasets with diverse underlying data, ABIDE and ADNI, for the prediction of Autism Spectrum Disorder and conversion to Alzheimer's disease, respectively. Our analysis shows that our novel framework can improve over state-of-the-art results on both databases, with 70.4% classification accuracy for ABIDE and 80.0% for ADNI.

Deep Gaussian processes (DGPs) provide a Bayesian non-parametric alternative to standard parametric deep learning models. A DGP is formed by stacking multiple GPs resulting in a well-regularized composition of functions. The Bayesian framework that equips the model with attractive properties, such as implicit capacity control and predictive uncertainty, makes it at the same time challenging to combine with a convolutional structure. This has hindered the application of DGPs in computer vision tasks, an area where deep parametric models (i.e. CNNs) have made breakthroughs. Standard kernels used in DGPs such as radial basis functions (RBFs) are insufficient for handling pixel variability in raw images. In this paper, we build on the recent convolutional GP to develop Convolutional DGP (CDGP) models which effectively capture image level features through the use of convolution kernels, therefore opening up the way for applying DGPs to computer vision tasks. Our model learns local spatial influence and outperforms strong GP based baselines on multi-class image classification. We also consider various constructions of convolution kernel over the image patches, analyze the computational trade-offs and provide an efficient framework for convolutional DGP models. The experimental results on image data such as MNIST, rectangles-image, CIFAR10 and Caltech101 demonstrate the effectiveness of the proposed approaches.

The generalization ability of a neural network depends on the optimization procedure used for training it. For practitioners and theoreticians, it is essential to identify which properties of the optimization procedure influence generalization. In this paper, we observe that prioritizing the training of distinct layers in a network significantly impacts its generalization ability, sometimes causing differences of up to 30% in test accuracy. In order to better monitor and control such prioritization, we propose to define layer-level training speed as the rotation rate of the layer's weight vector (denoted by layer rotation rate hereafter), and develop Layca, an optimization algorithm that enables direct control over it through each layer's learning rate parameter, without being affected by gradient propagation phenomena (e.g. vanishing gradients). We show that controlling layer rotation rates enables Layca to significantly outperform SGD with the same amount of learning rate tuning on three different tasks (up to 10% test error improvement). Furthermore, we provide experiments that suggest that several intriguing observations related to the training of deep models, i.e. the presence of plateaus in learning curves, the impact of weight decay, and the bad generalization properties of adaptive gradient methods, are all due to specific configurations of layer rotation rates. Overall, our work reveals that layer rotation rates are an important factor for generalization, and that monitoring it should be a key component of any deep learning experiment.

We present a personalized and reliable prediction model for healthcare, which can provide individually tailored medical services such as diagnosis, disease treatment and prevention. Our proposed framework targets to making reliable predictions from time-series data, such as Electronic Health Records (EHR), by modeling two complementary components: i) shared component that captures global trend across diverse patients and ii) patient-specific component that models idiosyncratic variability for each patient. To this end, we propose a composite model of a deep recurrent neural network (RNN) to exploit expressive power of the RNN in estimating global trends from large number of patients, and Gaussian Processes (GP) to probabilistically model individual time-series given relatively small number of time points. We evaluate the strength of our model on diverse and heterogeneous tasks in EHR datasets. The results show that our model significantly outperforms baselines such as RNN, demonstrating clear advantage over existing models when working with noisy medical data.

Clustering using neural networks has recently demon- strated promising performance in machine learning and computer vision applications. However, the performance of current approaches is limited either by unsupervised learn- ing or their dependence on large set of labeled data sam- ples. In this paper, we propose ClusterNet that uses pair- wise semantic constraints from very few labeled data sam- ples (< 5% of total data) and exploits the abundant un- labeled data to drive the clustering approach. We define a new loss function that uses pairwise semantic similarity between objects combined with constrained k-means clus- tering to efficiently utilize both labeled and unlabeled data in the same framework. The proposed network uses con- volution autoencoder to learn a latent representation that groups data into k specified clusters, while also learning the cluster centers simultaneously. We evaluate and com- pare the performance of ClusterNet on several datasets and state of the art deep clustering approaches.

We show that finite-width deep ReLU neural networks yield rate-distortion optimal approximation (B\"olcskei et al., 2018) of polynomials, windowed sinusoidal functions, one-dimensional oscillatory textures, and the Weierstrass function, a fractal function which is continuous but nowhere differentiable. Together with their recently established universal approximation property of affine function systems (B\"olcskei et al., 2018), this shows that deep neural networks approximate vastly different signal structures generated by the affine group, the Weyl-Heisenberg group, or through warping, and even certain fractals, all with approximation error decaying exponentially in the number of neurons. We also prove that in the approximation of sufficiently smooth functions finite-width deep networks require strictly smaller connectivity than finite-depth wide networks.

The objective of this work is to take advantage of deep neural networks in order to make next day crime count predictions in a fine-grain city partition. We make predictions using Chicago and Portland crime data, which is augmented with additional datasets covering weather, census data, and public transportation. The crime counts are broken into 10 bins and our model predicts the most likely bin for a each spatial region at a daily level. We train this data using increasingly complex neural network structures, including variations that are suited to the spatial and temporal aspects of the crime prediction problem. With our best model we are able to predict the correct bin for overall crime count with 75.6% and 65.3% accuracy for Chicago and Portland, respectively. The results show the efficacy of neural networks for the prediction problem and the value of using external datasets in addition to standard crime data.

Deep-neural-network (DNN) based noise suppression systems yield significant improvements over conventional approaches such as spectral subtraction and non-negative matrix factorization, but do not generalize well to noise conditions they were not trained for. In comparison to DNNs, humans show remarkable noise suppression capabilities that yield successful speech intelligibility under various adverse listening conditions and negative signal-to-noise ratios (SNRs). Motivated by the excellent human performance, this paper explores whether numerical models that simulate human cochlear signal processing can be combined with DNNs to improve the robustness of DNN based noise suppression systems. Five cochlear models were coupled to fully-connected and recurrent NN-based noise suppression systems and were trained and evaluated for a variety of noise conditions using objective metrics: perceptual speech quality (PESQ), segmental SNR and cepstral distance. The simulations show that biophysically-inspired cochlear models improve the generalizability of DNN-based noise suppression systems for unseen noise and negative SNRs. This approach thus leads to robust noise suppression systems that are less sensitive to the noise type and noise level. Because cochlear models capture the intrinsic nonlinearities and dynamics of peripheral auditory processing, it is shown here that accounting for their deterministic signal processing improves machine hearing and avoids overtraining of multi-layer DNNs. We hence conclude that machines hear better when realistic cochlear models are used at the input of DNNs.

Complexity manuscript No. (will be inserted by the editor) Abstract We study the use of feedforward neural networks (FNN) to develop models of nonlinear dynamical systems from data. Emphasis is placed on predictions at long times, with limited data availability. Inspired by global stability analysis, and the observation of strong correlation between the local error and the maximum singular value of the Jacobian of the ANN, we introduce Jacobian regularization in the loss function. This regularization suppresses the sensitivity of the prediction to the local error and is shown to improve accuracy and robustness. Comparison between the proposed approach and sparse polynomial regression is presented in numerical examples ranging from simple ODE systems to nonlinear PDE systems including vortex shedding behind a cylinder, and instability-driven buoyant mixing flow. Furthermore, limitations of feedforward neural networks are highlighted, especially when the training data does not include a low dimensional attractor. Strategies of data augmentation are presented as remedies to address these issues to a certain extent. Keywords data-driven modeling · artificial neural networks · Jacobian regularization · machine learning 1 Introduction The need to model dynamical behavior from data is pervasive across science and engineering. Applications are found in diverse domains such as in control systems [44], time series modeling [40], and describing the evolution of coherent structures [6]. While data-driven modeling of dynamical systems can be broadly classified as a special case of system identification [25], it is important to note certain distinguishing qualities: the learning process may be performed off-line; physical systems may involve very high dimensions; and the goal may involve the prediction of longtime behaviour from limited training data.