Sparse modeling for signal processing and machine learning has been at the focus of scientific research for over two decades. Among others, supervised sparsity-aware learning comprises two major paths paved by: a) discriminative methods and b) generative methods. The latter, more widely known as Bayesian methods, enable uncertainty evaluation w.r.t. the performed predictions. Furthermore, they can better exploit related prior information and naturally introduce robustness into the model, due to their unique capacity to marginalize out uncertainties related to the parameter estimates. Moreover, hyper-parameters associated with the adopted priors can be learnt via the training data. To implement sparsity-aware learning, the crucial point lies in the choice of the function regularizer for discriminative methods and the choice of the prior distribution for Bayesian learning. Over the last decade or so, due to the intense research on deep learning, emphasis has been put on discriminative techniques. However, a come back of Bayesian methods is taking place that sheds new light on the design of deep neural networks, which also establish firm links with Bayesian models and inspire new paths for unsupervised learning, such as Bayesian tensor decomposition. The goal of this article is two-fold. First, to review, in a unified way, some recent advances in incorporating sparsity-promoting priors into three highly popular data modeling tools, namely deep neural networks, Gaussian processes, and tensor decomposition. Second, to review their associated inference techniques from different aspects, including: evidence maximization via optimization and variational inference methods. Challenges such as small data dilemma, automatic model structure search, and natural prediction uncertainty evaluation are also discussed. Typical signal processing and machine learning tasks are demonstrated.

This paper addresses a data integration problem: given several mutually consistent datasets each of which measures a subset of the variables of interest, how can one construct a probabilistic model that fits the data and gives reasonable answers to questions which are under-determined by the data? Here we show how to obtain a Bayesian network model which represents the unique probability function that agrees with the probability distributions measured by the datasets and otherwise has maximum entropy. We provide a general algorithm, OBN-cDS, which offers substantial efficiency savings over the standard brute-force approach to determining the maximum entropy probability function. Furthermore, we develop modifications to the general algorithm which enable further efficiency savings but which are only applicable in particular situations. We show that there are circumstances in which one can obtain the model (i) directly from the data; (ii) by solving algebraic problems; and (iii) by solving relatively simple independent optimisation problems.

Some interviewers ask hard questions while others ask relatively easy questions. As an interviewee, it is your choice to go prepared. And when it comes to a domain like Machine Learning, preparations might fall short. You have to be prepared for everything. While preparing, you might have stuck at a point where you wonder what more shall I read. Well, based on almost 15-17 data science interviews that I have attended, here I have put 15, very commonly asked, as well as important Data Science and Machine Learning related questions that were asked to me in almost all of them and I recommend you must study these thoroughly.

Fragility hip fracture increases morbidity and mortality in older adult patients, especially within the first year. Identification of patients at high risk of death facilitates modification of associated perioperative factors that can reduce mortality. Various machine learning algorithms have been developed and are widely used in healthcare research, particularly for mortality prediction. This study aimed to develop and internally validate 7 machine learning models to predict 1-year mortality after fragility hip fracture. This retrospective study included patients with fragility hip fractures from a single center (Siriraj Hospital, Bangkok, Thailand) from July 2016 to October 2018. A total of 492 patients were enrolled. They were randomly categorized into a training group (344 cases, 70%) or a testing group (148 cases, 30%). Various machine learning techniques were used: the Gradient Boosting Classifier (GB), Random Forests Classifier (RF), Artificial Neural Network Classifier (ANN), Logistic Regression Classifier (LR), Naive Bayes Classifier (NB), Support Vector Machine Classifier (SVM), and K-Nearest Neighbors Classifier (KNN). All models were internally validated by evaluating their performance and the area under a receiver operating characteristic curve (AUC). For the testing dataset, the accuracies were GB model = 0.93, RF model = 0.95, ANN model = 0.94, LR model = 0.91, NB model = 0.89, SVM model = 0.90, and KNN model = 0.90. All models achieved high AUCs that ranged between 0.81 and 0.99. The RF model also provided a negative predictive value of 0.96, a positive predictive value of 0.93, a specificity of 0.99, and a sensitivity of 0.68. Our machine learning approach facilitated the successful development of an accurate model to predict 1-year mortality after fragility hip fracture. Several machine learning algorithms (eg, Gradient Boosting and Random Forest) had the potential to provide high predictive performance based on the clinical parameters of each patient. The web application is available at www.hipprediction.com . External validation in a larger group of patients or in different hospital settings is warranted to evaluate the clinical utility of this tool. Thai Clinical Trials Registry (22 February 2021; reg. no. TCTR20210222003 ).

When starting out with Data Science, there is so much to learn it can become quite overwhelming. This guide will help aspiring data scientists and machine learning engineers gain better knowledge and experience. I will list different types of machine learning algorithms, which can be used with both Python and R. Linear Regression is the simplest Machine learning algorithm that branches off from Supervised Learning. It is primarily used to solve regression problems and make predictions on continuous dependent variables with the knowledge from independent variables. The goal of Linear Regression is to find the line of best fit, which can help predict the output for continuous dependent variables.

Multi-defender Stackelberg Security Games (MSSG) have recently gained increasing attention in the literature. However, the solutions offered to date are highly sensitive, wherein even small perturbations in the attacker's utility or slight uncertainties thereof can dramatically change the defenders' resulting payoffs and alter the equilibrium. In this paper, we introduce a robust model for MSSGs, which admits solutions that are resistant to small perturbations or uncertainties in the game's parameters. First, we formally define the notion of robustness, as well as the robust MSSG model. Then, for the non-cooperative setting, we prove the existence of a robust approximate equilibrium in any such game, and provide an efficient construction thereof. For the cooperative setting, we show that any such game admits a robust approximate alpha-core, provide an efficient construction thereof, and prove that stronger types of the core may be empty. Interestingly, the robust solutions can substantially increase the defenders' utilities over those of the non-robust ones.

The Expectation-Maximization (EM) algorithm has been predominantly used to approximate the maximum likelihood estimation of the location-scale Gaussian mixtures. However, when the models are over-specified, namely, the chosen number of components to fit the data is larger than the unknown true number of components, EM needs a polynomial number of iterations in terms of the sample size to reach the final statistical radius; this is computationally expensive in practice. The slow convergence of EM is due to the missing of the locally strong convexity with respect to the location parameter on the negative population log-likelihood function, i.e., the limit of the negative sample log-likelihood function when the sample size goes to infinity. To efficiently explore the curvature of the negative log-likelihood functions, by specifically considering two-component location-scale Gaussian mixtures, we develop the Exponential Location Update (ELU) algorithm. The idea of the ELU algorithm is that we first obtain the exact optimal solution for the scale parameter and then perform an exponential step-size gradient descent for the location parameter. We demonstrate theoretically and empirically that the ELU iterates converge to the final statistical radius of the models after a logarithmic number of iterations. To the best of our knowledge, it resolves the long-standing open question in the literature about developing an optimization algorithm that has optimal statistical and computational complexities for solving parameter estimation even under some specific settings of the over-specified Gaussian mixture models.

Graph Neural Networks (GNN) provide a powerful framework that elegantly integrates Graph theory with Machine learning for modeling and analysis of networked data. We consider the problem of quantifying the uncertainty in predictions of GNN stemming from modeling errors and measurement uncertainty. We consider aleatoric uncertainty in the form of probabilistic links and noise in feature vector of nodes, while epistemic uncertainty is incorporated via a probability distribution over the model parameters. We propose a unified approach to treat both sources of uncertainty in a Bayesian framework, where Assumed Density Filtering is used to quantify aleatoric uncertainty and Monte Carlo dropout captures uncertainty in model parameters. Finally, the two sources of uncertainty are aggregated to estimate the total uncertainty in predictions of a GNN. Results in the real-world datasets demonstrate that the Bayesian model performs at par with a frequentist model and provides additional information about predictions uncertainty that are sensitive to uncertainties in the data and model.

Deep neural networks (DNNs) with noisy weights, which we refer to as noisy neural networks (NoisyNNs), arise from the training and inference of DNNs in the presence of noise. NoisyNNs emerge in many new applications, including the wireless transmission of DNNs, the efficient deployment or storage of DNNs in analog devices, and the truncation or quantization of DNN weights. This paper studies a fundamental problem of NoisyNNs: how to reconstruct the DNN weights from their noisy manifestations. While all prior works relied on the maximum likelihood (ML) estimation, this paper puts forth a denoising approach to reconstruct DNNs with the aim of maximizing the inference accuracy of the reconstructed models. The superiority of our denoiser is rigorously proven in two small-scale problems, wherein we consider a quadratic neural network function and a shallow feedforward neural network, respectively. When applied to advanced learning tasks with modern DNN architectures, our denoiser exhibits significantly better performance than the ML estimator. Consider the average test accuracy of the denoised DNN model versus the weight variance to noise power ratio (WNR) performance. When denoising a noisy ResNet34 model arising from noisy inference, our denoiser outperforms ML estimation by up to 4.1 dB to achieve a test accuracy of 60%.When denoising a noisy ResNet18 model arising from noisy training, our denoiser outperforms ML estimation by 13.4 dB and 8.3 dB to achieve test accuracies of 60% and 80%, respectively.

The ranking of experiments by expected information gain (EIG) in Bayesian experimental design is sensitive to changes in the model's prior distribution, and the approximation of EIG yielded by sampling will have errors similar to the use of a perturbed prior. We define and analyze \emph{robust expected information gain} (REIG), a modification of the objective in EIG maximization by minimizing an affine relaxation of EIG over an ambiguity set of distributions that are close to the original prior in KL-divergence. We show that, when combined with a sampling-based approach to estimating EIG, REIG corresponds to a `log-sum-exp' stabilization of the samples used to estimate EIG, meaning that it can be efficiently implemented in practice. Numerical tests combining REIG with variational nested Monte Carlo (VNMC), adaptive contrastive estimation (ACE) and mutual information neural estimation (MINE) suggest that in practice REIG also compensates for the variability of under-sampled estimators.