Linear regression is arguably the most prominent among statistical inference methods, popular both for its simplicity as well as its broad applicability. On par with data-intensive applications, the sheer size of linear regression problems creates an ever growing demand for quick and cost efficient solvers. Fortunately, a significant percentage of the data accrued can be omitted while maintaining a certain quality of statistical inference with an affordable computational budget. The present paper introduces means of identifying and omitting "less informative" observations in an online and data-adaptive fashion, built on principles of stochastic approximation and data censoring. First- and second-order stochastic approximation maximum likelihood-based algorithms for censored observations are developed for estimating the regression coefficients. Online algorithms are also put forth to reduce the overall complexity by adaptively performing censoring along with estimation. The novel algorithms entail simple closed-form updates, and have provable (non)asymptotic convergence guarantees. Furthermore, specific rules are investigated for tuning to desired censoring patterns and levels of dimensionality reduction. Simulated tests on real and synthetic datasets corroborate the efficacy of the proposed data-adaptive methods compared to data-agnostic random projection-based alternatives.

In this era of large-scale data, distributed systems built on top of clusters of commodity hardware provide cheap and reliable storage and scalable processing of massive data. Here, we review recent work on developing and implementing randomized matrix algorithms in large-scale parallel and distributed environments. Randomized algorithms for matrix problems have received a great deal of attention in recent years, thus far typically either in theory or in machine learning applications or with implementations on a single machine. Our main focus is on the underlying theory and practical implementation of random projection and random sampling algorithms for very large very overdetermined (i.e., overconstrained) $\ell_1$ and $\ell_2$ regression problems. Randomization can be used in one of two related ways: either to construct sub-sampled problems that can be solved, exactly or approximately, with traditional numerical methods; or to construct preconditioned versions of the original full problem that are easier to solve with traditional iterative algorithms. Theoretical results demonstrate that in near input-sparsity time and with only a few passes through the data one can obtain very strong relative-error approximate solutions, with high probability. Empirical results highlight the importance of various trade-offs (e.g., between the time to construct an embedding and the conditioning quality of the embedding, between the relative importance of computation versus communication, etc.) and demonstrate that $\ell_1$ and $\ell_2$ regression problems can be solved to low, medium, or high precision in existing distributed systems on up to terabyte-sized data.

We describe two nonconventional algorithms for linear regression, called GAME and CLASH. The salient characteristics of these approaches is that they exploit the convex $\ell_1$-ball and non-convex $\ell_0$-sparsity constraints jointly in sparse recovery. To establish the theoretical approximation guarantees of GAME and CLASH, we cover an interesting range of topics from game theory, convex and combinatorial optimization. We illustrate that these approaches lead to improved theoretical guarantees and empirical performance beyond convex and non-convex solvers alone.

Markov chain Monte Carlo (MCMC) is a popular tool for Bayesian inference.However, MCMC cannot be practically applied to large data sets because of theprohibitive cost of evaluating every likelihood term at every iteration. Here we present Firefly Monte Carlo (FlyMC) MCMC algorithm with auxiliary variables that only queries the likelihoods of a subset of the data at each iteration yet simulates from the exact posterior distribution. FlyMC is compatible with modern MCMC algorithms, and only requires a lower bound on the per-datum likelihood factors. In experiments, we find that FlyMC generates samples from the posterior more than an order of magnitude faster than regular MCMC, allowing MCMC methods to tackle larger datasets than were previously considered feasible.

Nowadays web users actively generate content on different social media platforms. The large number of users requiring personalized services creates a unique opportunity for researchers to explore user modelling. Substantial research has been done by utilizing user generated content to model users by applying different classification or regression techniques. These techniques are powerful types of machine learning approaches, however they only partially model social media users. In this work, we introduce a new statistical relational learning (SRL) framework suitable for this purpose, which we call PSL Q . PSL Q is the first SRL framework that supports reasoning with soft quantifiers, such as “most” and “a few”. Indeed, in models for social media it is common to assume that friends are influenced by each other’s behavior, beliefs, and preferences. Thus, having a trait only becomes probable once most or some of one’s friends have that trait. Expressing this dependency requires a soft quantifier, which can be modeled with PSL^Q. Our experimental results for link prediction in social trust networks demonstrate that the use of soft quantifiers not only allows for a natural and intuitive formulation of domain knowledge, but also improves the accuracy of inferred results.

Dynamic Programming (DP) over tree decompositions is a well-established method to solve problems — that are in general NP-hard — efficiently for instances of small treewidth. Experience shows that (i) heuristically computing a tree decomposition has negligible runtime compared to the DP step; (ii) DP algorithms exhibit a high variance in runtime when using different tree decompositions; in fact, given an instance of the problem at hand, even decompositions of the same width might yield extremely diverging runtimes. We thus propose here a novel and general method that is based on a selection of the best decomposition from an available pool of heuristically generated ones. For this purpose, we require machine learning techniques based on features of the decomposition rather than on the actual problem instance. We report on extensive experiments in different problem domains which show a significant speedup when choosing the tree decomposition according to this concept over simply using an arbitrary one of the same width.

Differential privacy preserving regression models guarantee protection against attempts to infer whether a subject was included in the training set used to derive a model. It is not designed to protect attribute privacy of a target individual when model inversion attacks are launched. In model inversion attacks, an adversary uses the released model to make predictions of sensitive attributes (used as input to the model) of a target individual when some background information about the target individual is available. Previous research showed that existing differential privacy mechanisms cannot effectively prevent model inversion attacks while retaining model efficacy. In this paper, we develop a novel approach which leverages the functional mechanism to perturb coefficients of the polynomial representation of the objective function but effectively balances the privacy budget for sensitive and non-sensitive attributes in learning the differential privacy preserving regression model. Theoretical analysis and empirical evaluations demonstrate our approach can effectively prevent model inversion attacks and retain model utility.

Domain adaptation aims at learning robust classifiers across domains using labeled data from a source domain. Representation learning methods, which project the original features to a new feature space, have been proved to be quite effective for this task. However, these unsupervised methods neglect the domain information of the input and are not specialized for the classification task. In this work, we address two key factors to guide the representation learning process for domain adaptation of sentiment classification — one is domain supervision, enforcing the learned representation to better predict the domain of an input, and the other is sentiment supervision which utilizes the source domain sentiment labels to learn sentiment-favorable representations. Experimental results show that these two factors significantly improve the proposed models as expected.

Recently, several Web-scale knowledge harvesting systems have been built, each of which is competent at extracting information from certain types of data (e.g., unstructured text, structured tables on the web, etc.). In order to determine the response to a new query posed to such systems (e.g., is sugar a healthy food?), it is useful to integrate opinions from multiple systems. If a response is desired within a specific time budget (e.g., in less than 2 seconds), then maybe only a subset of these resources can be queried. In this paper, we address the problem of knowledge integration for on-demand time-budgeted query answering. We propose a new method, AskWorld, which learns a policy that chooses which queries to send to which resources, by accommodating varying budget constraints that are available only at query (test) time. Through extensive experiments on real world datasets, we demonstrate AskWorld’s capability in selecting most informative resources to query within test-time constraints, resulting in improved performance compared to competitive baselines.

We study the problem of structured output learning from a regression perspective. We first provide a general formulation of the kernel dependency estimation (KDE) problem using operator-valued kernels. We show that some of the existing formulations of this problem are special cases of our framework. We then propose a covariance-based operator-valued kernel that allows us to take into account the structure of the kernel feature space. This kernel operates on the output space and encodes the interactions between the outputs without any reference to the input space. To address this issue, we introduce a variant of our KDE method based on the conditional covariance operator that in addition to the correlation between the outputs takes into account the effects of the input variables. Finally, we evaluate the performance of our KDE approach using both covariance and conditional covariance kernels on two structured output problems, and compare it to the state-of-the-art kernel-based structured output regression methods.