We consider regularized stochastic learning and online optimization problems, where the objective function is the sum of two convex terms: one is the loss function of the learning task, and the other is a simple regularization term such as ℓ1-norm for promoting sparsity. We develop extensions of Nesterov’s dual averaging method, that can exploit the regularization structure in an online setting. At each iteration of these methods, the learning variables are adjusted by solving a simple minimization problem that involves the running average of all past subgradients of the loss function and the whole regularization term, not just its subgradient. In the case of ℓ1-regularization, our method is particularly effective in obtaining sparse solutions. We show that these methods achieve the optimal convergence rates or regret bounds that are standard in the literature on stochastic and online convex optimization. For stochastic learning problems in which the loss functions have Lipschitz continuous gradients, we also present an accelerated version of the dual averaging method.

Conditional log-linear models are a commonly used method for structured prediction. Efficient learning of parameters in these models is therefore an important problem. This paper describes an exponentiated gradient (EG) algorithm for training such models. EG is applied to the convex dual of the maximum likelihood objective; this results in both sequential and parallel update algorithms, where in the sequential algorithm parameters are updated in an online fashion. We provide a convergence proof for both algorithms. Our analysis also simplifies previous results on EG for max-margin models, and leads to a tighter bound on convergence rates. Experiments on a large-scale parsing task show that the proposed algorithm converges much faster than conjugate-gradient and L-BFGS approaches both in terms of optimization objective and test error. 1.

Abstract not found

Abstract not found

Quite a bit is known about minimizing different kinds of regret in experts problems, and how these regret types relate to types of equilibria in the multiagent setting of repeated matrix games. Much less is known about the possible kinds of regret in online convex programming problems (OCPs), or about equilibria in the analogous multiagent setting of repeated convex games. This gap is unfortunate, since convex games are much more expressive than matrix games, and since many important machine learning problems can be expressed as OCPs. In this paper, we work to close this gap: we analyze a spectrum of regret types which lie between external and swap regret, along with their corresponding equilibria, which lie between coarse correlated and correlated equilibrium. We also analyze algorithms for minimizing these regret types. As examples of our framework, we derive algorithms for learning correlated equilibria in polyhedral convex games and extensive-form correlated equilibria in extensive-form games. The former is exponentially more efficient than previous algorithms, and the latter is the first of its type. 1.

We describe a primal-dual framework for the design and analysis of online strongly convex optimization algorithms. Our framework yields the tightest known logarithmic regret bounds for Follow-The-Leader and for the gradient descent algorithm proposed in Hazan et al. [2006]. We then show that one can interpolate between these two extreme cases. In particular, we derive a new algorithm that shares the computational simplicity of gradient descent but achieves lower regret in many practical situations. Finally, we further extend our framework for generalized strongly convex functions. 1

Boosting algorithms build highly accurate prediction mechanisms from a collection of lowaccuracy predictors. To do so, they employ the notion of weak-learnability. The starting point of this paper is a proof which shows that weak learnability is equivalent to linear separability with ℓ1 margin. While this equivalence is a direct consequence of von Neumann’s minimax theorem, we derive the equivalence directly using Fenchel duality. We then use our derivation to describe a family of relaxations to the weak-learnability assumption that readily translates to a family of relaxations of linear separability with margin. This alternative perspective sheds new light on known soft-margin boosting algorithms and also enables us to derive several new relaxations of the notion of linear separability. Last, we describe and analyze an efficient boosting framework that can be used for minimizing the loss functions derived from our family of relaxations. In particular, we obtain efficient boosting algorithms for maximizing hard and soft versions of the ℓ1 margin.

Online convex programming has recently emerged as a powerful primitive for designing machine learning algorithms. For example, OCP can be used for learning a linear classifier, dynamically rebalancing a binary search tree, finding the shortest path in a graph with unknown edge lengths, solving a structured classification problem, or finding a good strategy in an extensive-form game. Several researchers have designed no-regret algorithms for OCP. But, compared to algorithms for special cases of OCP such as learning from expert advice, these algorithms are not very numerous or flexible. In learning from expert advice, one tool which has proved particularly valuable is the correspondence between no-regret algorithms and convex potential functions: by reasoning about these potential functions, researchers have designed algorithms with a wide variety of useful guarantees such as good performance when the target hypothesis is sparse. Until now, there has been no such recipe for the more general OCP problem, and therefore no ability to tune OCP algorithms to take advantage of properties of the problem or data. In this paper we derive a new class of no-regret learning algorithms for OCP. These Lagrangian Hedging algorithms are based on a general class of potential functions, and are a direct generalization of known learning rules like weighted majority and external-regret matching. In addition to proving regret bounds, we demonstrate our algorithms learning to play one-card poker. 1

We develop a theory of online learning by defining several complexity measures. Among them are analogues of Rademacher complexity, covering numbers and fatshattering dimension from statistical learning theory. Relationship among these complexity measures, their connection to online learning, and tools for bounding them are provided. We apply these results to various learning problems. We provide a complete characterization of online learnability in the supervised setting. 1

Label ranking is the task of ordering labels with respect to their relevance to an input instance. We describe a unified approach for the online label ranking task. We do so by casting the online learning problem as a game against a competitor who receives all the examples in advance and sets its label ranker to be the optimal solution of a constrained optimization problem. This optimization problem consists of two terms: the empirical label-ranking loss of the competitor and a complexity measure of the competitor’s ranking function. We then describe and analyze a framework for online label ranking that incrementally ascends the dual problem corresponding to the competitor’s optimization problem. The generality of our framework enables us to derive new online update schemes. In particular, we use the relative entropy as a complexity measure to derive efficient multiplicative algorithms for the label ranking task. Depending on the specific form of the instances, the multiplicative updates either have a closed form or can be calculated very efficiently by tailoring an interior point procedure to the label ranking task. We demonstrate the potential of our approach in a few experiments with email categorization tasks. 1