Indefinite kernel support vector machine (IKSVM) has recently attracted increasing attentions in machine learning. Different from traditional SVMs, IKSVM essentially is a non-convex optimization problem. Some algorithms directly change the spectrum of the indefinite kernel matrix at the cost of losing some valuable information involved in the kernels so as to transform the non-convex problem into a convex one. Other algorithms aim to solve the dual form of IKSVM, but suffer from the dual gap between the primal and dual problems in the case of indefinite kernels. In this paper, we directly focus on the non-convex primal form of IKSVM and propose a novel algorithm termed as IKSVM-DC. According to the characteristics of the spectrum for the indefinite kernel matrix, IKSVM-DC decomposes the objective function into the subtraction of two convex functions and thus reformulates the primal problem as a difference of convex functions (DC) programming which can be optimized by the DC algorithm (DCA). In order to accelerate convergence rate, IKSVM-DC further combines the classical DCA with a line search step along the descent direction at each iteration. A theoretical analysis is then presented to validate that IKSVM-DC can converge to a local minimum. Systematical experiments on real-world datasets demonstrate the superiority of IKSVM-DC compared to state-of-the-art IKSVM related algorithms.

A number of machine learning domains,such as information retrieval, recommender systems, kernel learning, neural network-biological systems etc,deal with importance scores. Very often, there existsome prior knowledge that could help improve the performance.In many cases, these prior knowledge manifest themselves in the rank ordering constraints.These inequality constraints are usually very difficult to deal with in optimization.In this paper, we provide a slack variable transformation methods, which effectively eliminatesthe rank ordering inequality constraints, and thus simplify the learning task significantly.We apply this transformation in kernel learning problem, and also provide an efficient algorithm tosolved the transformed system. On seven datasets,our approach reduces the computational time by orders of magnitudes as compared to the current standardquadratically constrained quadratic programming(QCQP) optimization approach.

Multi-Armed Bandit (MAB) framework has been successfully applied in many web applications. However, many complex real-world applications that involve multiple content recommendations cannot fit into the traditional MAB setting. To address this issue, we consider an ordered combinatorial semi-bandit problem where the learner recommends S actions from a base set of K actions, and displays the results in S (out of M ) different positions. The aim is to maximize the cumulative reward with respect to the best possible subset and positions in hindsight. By the adaptation of a minimum-cost maximum-flow network, a practical algorithm based on Thompson sampling is derived for the (contextual) combinatorial problem, thus resolving the problem of computational intractability.With its potential to work with whole-page recommendation and any probabilistic models, to illustrate the effectiveness of our method, we focus on Gaussian process optimization and a contextual setting where click-through rate is predicted using logistic regression. We demonstrate the algorithms’ performance on synthetic Gaussian process problems and on large-scale news article recommendation datasets from Yahoo! Front Page Today Module.

The explosion of streaming data poses challenges to feature learning methods including linear discriminant analysis (LDA). Many existing LDA algorithms are not efficient enough to incrementally update with samples that sequentially arrive in various manners. First, we propose a new fast batch LDA (FLDA/QR) learning algorithm that uses the cluster centers to solve a lower triangular system that is optimized by the Cholesky-factorization. To take advantage of the intrinsically incremental mechanism of the matrix, we further develop an exact incremental algorithm (IFLDA/QR). The Gram-Schmidt process with reorthogonalization in IFLDA/QR significantly saves the space and time expenses compared with the rank-one QR-updating of most existing methods. IFLDA/QR is able to handle streaming data containing 1) new labeled samples in the existing classes, 2) samples of an entirely new (novel) class, and more significantly, 3) a chunk of examples mixed with those in 1) and 2). Both theoretical analysis and numerical experiments have demonstrated much lower space and time costs (2~10 times faster) than the state of the art, with comparable classification accuracy.

Existing methods mostly rely on distances (or similarities) In many fields of science and engineering, one is often to model the intrinsic structure of data. They either provide confronted with the problem of dimensionality reduction a similarity matrix as a prior (Belkin and Niyogi 2001; (Burges 2009; Van der Maaten, Postma, and van den Herik Schölkopf, Smola, and Muller 1999), or learn a similarity 2009). The problem aims to extract low-dimensional structures measurement based on a subset of distances in a local from high-dimensional datasets, which are generally region (Elhamifar and Vidal 2011; Saul and Roweis characterized by much fewer degrees of freedom than actual 2003), or directly learn a kernel matrix from data (Weinberger, number of features. Packer, and Saul 2005; Xiao, Sun, and Boyd 2006; In this paper, we are particularly interested in unveiling a Mao and Tsang 2010). These distances become unreliable if smooth skeleton structure in a latent space from data with the data is noisy. Moreover, they lack the ability to model a noise. Figure 1 illustrates an intuitive example in which synthetic smooth skeleton from noisy data. As shown in Figure 1, the data points are drawn from a smooth circle with noises strict distance preservation in maximum variance unfolding in two-dimensional space. It is challenging to recover the (MVU) (Weinberger, Sha, and Saul 2004) fails to capture the circle (Figures 1(c) and 1(d)) from the noisy data without smooth circle from the data (see Figure 1 (b)).

We perform online interactive recommendation via a factorization-based bandit algorithm. Low-rank matrix completion is performed over an incrementally constructed user-item preference matrix, where an upper confidence bound based item selection strategy is developed to balance the exploit/explore trade-off during online learning. Observable contextual features and dependency among users (e.g., social influence) are leveraged to improve the algorithm's convergence rate and help conquer cold-start in recommendation. A high probability sublinear upper regret bound is proved for the developed algorithm, where considerable regret reduction is achieved on both user and item sides. Extensive experimentations on both simulations and large-scale real-world datasets confirmed the advantages of the proposed algorithm compared with several state-of-the-art factorization-based and bandit-based collaborative filtering methods.

Direct contextual policy search methods learn to improve policy parameters and simultaneously generalize these parameters to different context or task variables. However, learning from high-dimensional context variables, such as camera images, is still a prominent problem in many real-world tasks. A naive application of unsupervised dimensionality reduction methods to the context variables, such as principal component analysis, is insufficient as task-relevant input may be ignored. In this paper, we propose a contextual policy search method in the model-based relative entropy stochastic search framework with integrated dimensionality reduction. We learn a model of the reward that is locally quadratic in both the policy parameters and the context variables. Furthermore, we perform supervised linear dimensionality reduction on the context variables by nuclear norm regularization. The experimental results show that the proposed method outperforms naive dimensionality reduction via principal component analysis and a state-of-the-art contextual policy search method.

We study the problem of cross-domain ranking, which addresses learning to rank objects from multiple interrelated domains. In many applications, we may have multiple interrelated domains, some of them with a large amount of training data and others with very little. We often wish to utilize the training data from all these related domains to help improve ranking performance. In this paper, we present a unified model: BayCDR for cross-domain ranking. BayCDR uses a latent space to measure the correlation between different domains, and learns the ranking functions from the interrelated domains via the latent space by a Bayesian model, where each ranking function is based on a weighted average model. An efficient learning algorithm based on variational inference and a generalization bound has been developed. To scale up to handle real large data, we also present a learning algorithm under the Map-Reduce programming model. Finally, we demonstrate the effectiveness and efficiency of BayCDR on large datasets.

Word2Vec recently popularized dense vector word representations as fixed-length features for machine learning algorithms and is in widespread use today. In this paper we investigate one of its core components, Negative Sampling, and propose efficient distributed algorithms that allow us to scale to vocabulary sizes of more than 1 billion unique words and corpus sizes of more than 1 trillion words.

Embedded feature selection is effective when both prediction and interpretation are needed. The Lasso and its extensions are standard methods for selecting a subset of features while optimizing a prediction function. In this paper, we are interested in embedded feature selection for multidimensional data, wherein (1) there is no need to reshape the multidimensional data into vectors and (2) structural information from multiple dimensions are taken into account. Our main contribution is a new method called Regularized multilinear regression and selection (Remurs) for automatically selecting a subset of features while optimizing prediction for multidimensional data. Both nuclear norm and the ℓ 1 -norm are carefully incorporated to derive a multi-block optimization algorithm with proved convergence. In particular, Remurs is motivated by fMRI analysis where the data are multidimensional and it is important to find the connections of raw brain voxels with functional activities. Experiments on synthetic and real data show the advantages of Remurs compared to Lasso, Elastic Net, and their multilinear extensions.