In this paper, it is shown how to extract a hypothesis with small risk from the ensemble of hypotheses generated by an arbitrary on-line learning algorithm run on an independent and identically distributed (i.i.d.) sample of data. Using a simple large deviation argument, we prove tight data-dependent bounds for the risk of this hypothesis in terms of an easily computable statistic associated with the on-line performance of the ensemble. Via sharp pointwise bounds on , we then obtain risk tail bounds for kernel Perceptron algorithms in terms of the spectrum of the empirical kernel matrix. These bounds reveal that the linear hypotheses found via our approach achieve optimal tradeoffs between hinge loss and margin size over the class of all linear functions, an issue that was left open by previous results.A distinctive feature of our approach is that the key tools for our analysis come from the model of prediction of individual sequences; i.e., a model making no probabilistic assumptions on the source generating the data. In fact, these tools turn out to be so powerful that we only need very elementary statistical facts to obtain our final risk bounds.Index Terms-Kernel functions, on-line learning, pattern recognition, perceptron algorithm, statistical learning theory.

We show that the convex envelope of the objective function of Mixed-Integer Programming problems with a specific structure is the perspective function of the continuous part of the objective function. Using a characterization of the subdifferential of the perspective function, we derive "perspective cuts", a family of valid inequalities for the problem. Perspective cuts can be shown to belong to the general family of disjunctive cuts, but they do not require the solution of a potentially costly nonlinear programming problem to be separated. Using perspective cuts substantially improves the performance of Branch-and-Cut approaches for at least two models that, either "naturally" or after a proper reformulation, have the required structure: the Unit Commitment problem in electrical power production and the Mean-Variance problem in portfolio optimization

We study on-line learning in the linear regression framework. Most of the performance bounds for on-line algorithms in this framework assume a constant learning rate. To achieve these bounds the learning rate must be optimized based on a posteriori information. This information depends on the whole sequence of examples and thus it is not available to any strictly on-line algorithm. We introduce new techniques for adaptively tuning the learning rate as the data sequence is progressively revealed. Our techniques allow us to prove essentially the same bounds as if we knew the optimal learning rate in advance. Moreover, such techniques apply to a wide class of on-line algorithms, including p-norm algorithms for generalized linear regression and Weighted Majority for linear regression with absolute loss. Our adaptive tunings are radically different from previous techniques, such as the so-called doubling trick. Whereas the doubling trick restarts the on-line algorithm several times using a constant learning rate for each run, our methods save information by changing the value of the learning rate very smoothly. In fact, for Weighted Majority over a finite set of experts our analysis provides a better leading constant than the doubling trick. © 2002 Elsevier Science (USA)

Classical collaborative filtering, and content-based filtering methods try to learn a static recommendation model given training data. These approaches are far from ideal in highly dynamic recommendation domains such as news recommendation and computational advertisement, where the set of items and users is very fluid. In this work, we investigate an adaptive clustering technique for content recommendation based on exploration-exploitation strategies in contextual multi-armed bandit settings. Our algorithm takes into account the collaborative effects that arise due to the interaction of the users with the items, by dynamically grouping users based on the items under consideration and, at the same time, grouping items based on the similarity of the clusterings induced over the users. The resulting algorithm thus takes advantage of preference patterns in the data in a way akin to collaborative filtering methods. We provide an empirical analysis on medium-size real-world datasets, showing scalability and increased prediction performance (as measured by click-through rate) over state-of-the-art methods for clustering bandits. We also provide a regret analysis within a standard linear stochastic noise setting.

The short-term Unit Commitment (UC) problem in hydro-thermal power generation is a largescale, Mixed-Integer NonLinear Program (MINLP), which is difficult to solve efficiently, especially for large-scale instances. It is possible to approximate the nonlinear objective function of the problem by means of piecewise-linear functions, so that UC can be approximated by a Mixed-Integer Linear Program (MILP); applying the available efficient general-purpose MILP solvers to the resulting formulations, good quality solutions can be obtained in a relatively short amount of time. We build on this approach, presenting a novel way to approximating the nonlinear objective function based on a recently developed class of valid inequalities for the problem, called "Perspective Cuts". At least for many realistic instances of a general basic formulation of UC, a MILP-based heuristic obtains comparable or slightly better solutions in less time when employing the new approach rather than the standard piecewise linearizations, while being not more difficult to implement and use. Furthermore, "dynamic" formulations, whereby the approximation is iteratively improved, provide even better results if the approximation is appropriately controlled.Key words: Hydro-Thermal Unit Commitment, Mixed-Integer Linear Program Formulations, Valid Inequalities. 3. NomenclatureThe notation used throughout the paper is stated below. For unit consistency, note that hourly intervals are considered. Constants: n number of time intervals (hours) T set of all time periods P set of thermal units H set of hydro cascades, each comprising one or more basin units H(h) set of individual hydro units cascade h ∈ H is composed of B(j) set of the immediate predecessors of hydro unit j t kj water time delay from plant k ∈ B(j) to the basin feeding hydro unit j

We consider two on-line learning frameworks: binary classification through linear threshold functions and linear regression. We study a family of on-line algorithms, called p-norm algorithms, introduced by Grove, Littlestone and Schuurmans in the context of deterministic binary classification. We show how to adapt these algorithms for use in the regression setting, and prove worst-case bounds on the square loss, using a technique from Kivinen and Warmuth. As pointed out by Grove, et al., these algorithms can be made to approach a version of the classification algorithm Winnow as p goes to infinity; similarly they can be made to approach the corresponding regression algorithm EG in the limit. Winnow and EG are notable for having loss bounds that grow only logarithmically in the dimension of the instance space. Here we describe another way to use the p-norm algorithms to achieve this logarithmic behavior. With the way to use them that we propose, it is less critical than with Winnow and EG to retune the parameters of the algorithm as the learning task changes. Since the correct setting of the parameters depends on characteristics of the learning task that are not typically known a priori by the learner, this gives the p-norm algorithms a desireable robustness. Our elaborations yield various new loss bounds in these on-line settings. Some of these bounds improve or generalize known results. Others are incomparable.

We show a regret minimization algorithm for setting the reserve price in second-price auctions. We make the assumption that all bidders draw their bids from the same unknown and arbitrary distribution. Our algorithm is computationally efficient, and achieves a regret of e O( √ T ), even when the number of bidders is stochastic with a known distribution.

We present a new approach, requiring the solution of a SemiDefinite Program, for decomposing the Hessian of a nonseparable Mixed-Integer Quadratic problem to permit using perspective cuts to improve its continuous relaxation bound. The new method favorably compares with a previously proposed one requiring a minimum eigenvalue computation.