Logistic regression with ℓ1 regularization has been proposed as a promising method for feature selection in classification problems. In this paper we describe an efficient interior-point method for solving large-scale ℓ1-regularized logistic regression problems. Small problems with up to a thousand or so features and examples can be solved in seconds on a PC; medium sized problems, with tens of thousands of features and examples, can be solved in tens of seconds (assuming some sparsity in the data). A variation on the basic method, that uses a preconditioned conjugate gradient method to compute the search step, can solve very large problems, with a million features and examples (e.g., the 20 Newsgroups data set), in a few minutes, on a PC. Using warm-start techniques, a good approximation of the entire regularization path can be computed much more efficiently than by solving a family of problems independently.

Markov networks are widely used in a wide variety of applications, in problems ranging from computer vision, to natural language, to computational biology. In most current applications, even those that rely heavily on learned models, the structure of the Markov network is constructed by hand, due to the lack of effective algorithms for learning Markov network structure from data. In this paper, we provide a computationally effective method for learning Markov network structure from data. Our method is based on the use of L1 regularization on the weights of the log-linear model, which has the effect of biasing the model towards solutions where many of the parameters are zero. This formulation converts the Markov network learning problem into a convex optimization problem in a continuous space, which can be solved using efficient gradient methods. A key issue in this setting is the (unavoidable) use of approximate inference, which can lead to errors in the gradient computation when the network structure is dense. Thus, we explore the use of different feature introduction schemes and compare their performance. We provide results for our method on synthetic data, and on two real world data sets: modeling the joint distribution of pixel values in the MNIST data, and modeling the joint distribution of genetic sequence variations in the human HapMap data. We show that our L1-based method achieves considerably higher generalization performance than the more standard L2-based method (a Gaussian parameter prior) or pure maximum-likelihood learning. We also show that we can learn MRF network structure at a computational cost that is not much greater than learning parameters alone, demonstrating the existence of a feasible method for this important problem. 1

The l-bfgs limited-memory quasi-Newton method is the algorithm of choice for optimizing the parameters of large-scale log-linear models with L2 regularization, but it cannot be used for an L1-regularized loss due to its non-differentiability whenever some parameter is zero. Efficient algorithms have been proposed for this task, but they are impractical when the number of parameters is very large. We present an algorithm Orthant-Wise Limited-memory Quasi-Newton (owlqn), based on l-bfgs, that can efficiently optimize the L1-regularized log-likelihood of log-linear models with millions of parameters. In our experiments on a parse reranking task, our algorithm was several orders of magnitude faster than an alternative algorithm, and substantially faster than lbfgs on the analogous L2-regularized problem. We also present a proof that owl-qn is guaranteed to converge to a globally optimal parameter vector. 1.

The key idea behind active learning is that a machine learning algorithm can achieve greater accuracy with fewer labeled training instances if it is allowed to choose the data from which is learns. An active learner may ask queries in the form of unlabeled instances to be labeled by an oracle (e.g., a human annotator). Active learning is well-motivated in many modern machine learning problems, where unlabeled data may be abundant but labels are difficult, time-consuming, or expensive to obtain. This report provides a general introduction to active learning and a survey of the literature. This includes a discussion of the scenarios in which queries can be formulated, and an overview of the query strategy frameworks proposed in the literature to date. An analysis of the empirical and theoretical evidence for active learning, a summary of several problem setting variants, and a discussion

Abstract. We consider the problem of estimating an unknown probability distribution from samples using the principle of maximum entropy (maxent). To alleviate overfitting with a very large number of features, we propose applying the maxent principle with relaxed constraints on the expectations of the features. By convex duality, this turns out to be equivalent to finding the Gibbs distribution minimizing a regularized version of the empirical log loss. We prove nonasymptotic bounds showing that, with respect to the true underlying distribution, this relaxed version of maxent produces density estimates that are almost as good as the best possible. These bounds are in terms of the deviation of the feature empirical averages relative to their true expectations, a number that can be bounded using standard uniform-convergence techniques. In particular, this leads to bounds that drop quickly with the number of samples, and that depend very moderately on the number or complexity of the features. We also derive and prove convergence for both sequential-update and parallel-update algorithms. Finally, we briefly describe experiments on data relevant to the modeling of species geographical distributions. 1

L1 regularized logistic regression is now a workhorse of machine learning: it is widely used for many classification problems, particularly ones with many features. L1 regularized logistic regression requires solving a convex optimization problem. However, standard algorithms for solving convex optimization problems do not scale well enough to handle the large datasets encountered in many practical settings. In this paper, we propose an efficient algorithm for L1 regularized logistic regression. Our algorithm iteratively approximates the objective function by a quadratic approximation at the current point, while maintaining the L1 constraint. In each iteration, it uses the efficient LARS (Least Angle Regression) algorithm to solve the resulting L1 constrained quadratic optimization problem. Our theoretical results show that our algorithm is guaranteed to converge to the global optimum. Our experiments show that our algorithm significantly outperforms standard algorithms for solving convex optimization problems. Moreover, our algorithm outperforms four previously published algorithms that were specifically designed to solve the L1 regularized logistic regression problem.

This thesis is about estimating probabilistic models to uncover useful hidden structure in data; specifically, we address the problem of discovering syntactic structure in natural language text. We present three new parameter estimation techniques that generalize the standard approach, maximum likelihood estimation, in different ways. Contrastive estimation maximizes the conditional probability of the observed data given a “neighborhood” of implicit negative examples. Skewed deterministic annealing locally maximizes likelihood using a cautious parameter search strategy that starts with an easier optimization problem than likelihood, and iteratively moves to harder problems, culminating in likelihood. Structural annealing is similar, but starts with a heavy bias toward simple syntactic structures and gradually relaxes the bias. Our estimation methods do not make use of annotated examples. We consider their performance in both an unsupervised model selection setting, where models trained under different initialization and regularization settings are compared by evaluating the training objective on a small set of unseen, unannotated development data, and supervised model selection, where the most accurate model on the development set (now with annotations)

Abstract. We present a unified and complete account of maximum entropy distribution estimation subject to constraints represented by convex potential functions or, alternatively, by convex regularization. We provide fully general performance guarantees and an algorithm with a complete convergence proof. As special cases, we can easily derive performance guarantees for many known regularization types, including ℓ1, ℓ2, ℓ 2 2 and ℓ1 + ℓ 2 2 style regularization. Furthermore, our general approach enables us to use information about the structure of the feature space or about sample selection bias to derive entirely new regularization functions with superior guarantees. We propose an algorithm solving a large and general subclass of generalized maxent problems, including all discussed in the paper, and prove its convergence. Our approach generalizes techniques based on information geometry and Bregman divergences as well as those based more directly on compactness. 1

We study the problem of modeling species geographic distributions, a critical problem in conservation biology. We propose the use of maximum-entropy techniques for this problem, specifically, sequential-update algorithms that can handle a very large number of features. We describe experiments comparing maxent with a standard distribution-modeling tool, called GARP, on a dataset containing observation data for North American breeding birds. We also study how well maxent performs as a function of the number of training examples and training time, analyze the use of regularization to avoid overfitting when the number of examples is small, and explore the interpretability of models constructed using maxent. 1.

We present a unified and complete account of maximum entropy density estimation subject to constraints represented by convex potential functions or, alternatively, by convex regularization. We provide fully general performance guarantees and an algorithm with a complete convergence proof. As special cases, we easily derive performance guarantees for many known regularization types, including ℓ1, ℓ2, ℓ 2 2, and ℓ1+ ℓ 2 2 style regularization. We propose an algorithm solving a large and general subclass of generalized maximum entropy problems, including all discussed in the paper, and prove its convergence. Our approach generalizes and unifies techniques based on information geometry and Bregman divergences as well as those based more directly on compactness. Our work is motivated by a novel application of maximum entropy to species distribution modeling, an important problem in conservation biology and ecology. In a set of experiments on real-world data, we demonstrate the utility of maximum entropy in this setting. We explore effects of different feature types, sample sizes, and regularization levels on the performance of maxent, and discuss interpretability of the resulting models.