High dimensional regression benefits from sparsity promoting regularizations. Screening rules leverage the known sparsity of the solution by ignoring some variables in the optimization, hence speeding up solvers. When the procedure is proven not to discard features wrongly the rules are said to be \emph{safe}. In this paper we derive new safe rules for generalized linear models regularized with $\ell_1$ and $\ell_1/\ell_2$ norms. The rules are based on duality gap computations and spherical safe regions whose diameters converge to zero. This allows to discard safely more variables, in particular for low regularization parameters. The GAP Safe rule can cope with any iterative solver and we illustrate its performance on coordinate descent for multi-task Lasso, binary and multinomial logistic regression, demonstrating significant speed ups on all tested datasets with respect to previous safe rules.

The random forest algorithm, proposed by L. Breiman in 2001, has been extremely successful as a general-purpose classification and regression method. The approach, which combines several randomized decision trees and aggregates their predictions by averaging, has shown excellent performance in settings where the number of variables is much larger than the number of observations. Moreover, it is versatile enough to be applied to large-scale problems, is easily adapted to various ad-hoc learning tasks, and returns measures of variable importance. The present article reviews the most recent theoretical and methodological developments for random forests. Emphasis is placed on the mathematical forces driving the algorithm, with special attention given to the selection of parameters, the resampling mechanism, and variable importance measures. This review is intended to provide non-experts easy access to the main ideas.

Learning accurate probabilistic models from data is crucial in many practical tasks in data mining. In this paper we present a new non-parametric calibration method called \textit{ensemble of near isotonic regression} (ENIR). The method can be considered as an extension of BBQ, a recently proposed calibration method, as well as the commonly used calibration method based on isotonic regression. ENIR is designed to address the key limitation of isotonic regression which is the monotonicity assumption of the predictions. Similar to BBQ, the method post-processes the output of a binary classifier to obtain calibrated probabilities. Thus it can be combined with many existing classification models. We demonstrate the performance of ENIR on synthetic and real datasets for the commonly used binary classification models. Experimental results show that the method outperforms several common binary classifier calibration methods. In particular on the real data, ENIR commonly performs statistically significantly better than the other methods, and never worse. It is able to improve the calibration power of classifiers, while retaining their discrimination power. The method is also computationally tractable for large scale datasets, as it is $O(N \log N)$ time, where $N$ is the number of samples.

We study parameter estimation and asymptotic inference for sparse nonlinear regression. More specifically, we assume the data are given by $y = f( x^\top \beta^* ) + \epsilon$, where $f$ is nonlinear. To recover $\beta^*$, we propose an $\ell_1$-regularized least-squares estimator. Unlike classical linear regression, the corresponding optimization problem is nonconvex because of the nonlinearity of $f$. In spite of the nonconvexity, we prove that under mild conditions, every stationary point of the objective enjoys an optimal statistical rate of convergence. In addition, we provide an efficient algorithm that provably converges to a stationary point. We also access the uncertainty of the obtained estimator. Specifically, based on any stationary point of the objective, we construct valid hypothesis tests and confidence intervals for the low dimensional components of the high-dimensional parameter $\beta^*$. Detailed numerical results are provided to back up our theory.

We consider the scenario where the parameters of a probabilistic model are expected to vary over time. We construct a novel prior distribution that promotes sparsity and adapts the strength of correlation between parameters at successive timesteps, based on the data. We derive approximate variational inference procedures for learning and prediction with this prior. We test the approach on two tasks: forecasting financial quantities from relevant text, and modeling language contingent on time-varying financial measurements.

Recently there has been substantial interest in spectral methods for learning dynamical systems. These methods are popular since they often offer a good tradeoff between computational and statistical efficiency. Unfortunately, they can be difficult to use and extend in practice: e.g., they can make it difficult to incorporate prior information such as sparsity or structure. To address this problem, we present a new view of dynamical system learning: we show how to learn dynamical systems by solving a sequence of ordinary supervised learning problems, thereby allowing users to incorporate prior knowledge via standard techniques such as L1 regularization. Many existing spectral methods are special cases of this new framework, using linear regression as the supervised learner. We demonstrate the effectiveness of our framework by showing examples where nonlinear regression or lasso let us learn better state representations than plain linear regression does; the correctness of these instances follows directly from our general analysis.

There has been much work on player modeling using game behavioral data collected. Many of the previous research projects that targeted this goal used aggregate game statistics as features to develop behavior models using both statistical and machine learning techniques. While existing methods have already led to interesting findings, we suspect that aggregated features discard valuable information such as temporal or sequential patterns, which may be important in deciphering information about decisionmaking, problem solving, or individual differences. Such sequential information is critical to analyze player behaviors especially in role-playing games (RPG) where players can face ample choices, experience different contexts, behave freely with individual propensities but possibly end up with similar aggregated statistics (e.g., levels, time spent). In this paper we intend to develop and apply a modeling technique that takes into consideration sequential patters to decipher individual differences in playing a Role Playing Game (RPG) game. Using an RPG with multiple affordances, we designed an experiment collecting granular in-game behaviors of 64 players. Using closed sequential pattern mining and logistic regression, we developed a model that uses gameplay action sequences to predict the real world characteristics, including gender, game play expertise and five personality traits (as defined by psychology). The results show that game expertise is a dominant factor that impacts in-game behaviors. The contribution of this paper is the algorithms we developed combined with a validation procedure to determine the reliability and validity of the results and the results themselves.

Mobile digital games are dominantly released under the freemium business model, but only a small fraction of the players makes any purchases. The ability to predict who will make a purchase enables optimization of marketing efforts, and tailoring customer relationship management to the specific user's profile. Here this challenge is addressed via two models for predicting purchasing players, using a 100,000 player dataset: 1) A classification model focused on predicting whether a purchase will occur or not. 2) a regression model focused on predicting the number of purchases a user will make. Both models are presented within a decision and regression tree framework for building rules that are actionable by companies. To the best of our knowledge, this is the first study investigating purchase decisions in freemium mobile products from a user behavior perspective and adopting behavior-driven learning approaches to this problem.

The parametric complexity is the key quantity in the minimum description length (MDL) approach to statistical model selection. Rissanen and others have shown that the parametric complexity of a statistical model approaches a simple function of the Fisher information volume of the model as the sample size $n$ goes to infinity. This paper derives higher-order asymptotic expansions for the parametric complexity, in the case of exponential families and independent and identically distributed data. These higher-order approximations are calculated for some examples and are shown to have better finite-sample behaviour than Rissanen's approximation. The higher-order terms are given as expressions involving cumulants (or, more naturally, the Amari-Chentsov tensors), and these terms are likely to be interesting in themselves since they arise naturally from the general information-theoretic principles underpinning MDL. The derivation given here specializes to an alternative and arguably simpler proof of Rissanen's result (for the case considered here), proving for the first time that his approximation is $O(n^{-1})$.

Frank-Wolfe (FW) algorithms have been often proposed over the last few years as efficient solvers for a variety of optimization problems arising in the field of Machine Learning. The ability to work with cheap projection-free iterations and the incremental nature of the method make FW a very effective choice for many large-scale problems where computing a sparse model is desirable. In this paper, we present a high-performance implementation of the FW method tailored to solve large-scale Lasso regression problems, based on a randomized iteration, and prove that the convergence guarantees of the standard FW method are preserved in the stochastic setting. We show experimentally that our algorithm outperforms several existing state of the art methods, including the Coordinate Descent algorithm by Friedman et al. (one of the fastest known Lasso solvers), on several benchmark datasets with a very large number of features, without sacrificing the accuracy of the model. Our results illustrate that the algorithm is able to generate the complete regularization path on problems of size up to four million variables in less than one minute.