Search

The problem of learning the kernel function with linear combinations of multiple kernels has attracted considerable attention recently in machine learning. Specially, by imposing an lp-norm penalty on the kernel combination coefficient, multiple kernel learning (MKL) was proved useful and effective for theoretical analysis and practical applications (Kloft et al., 2009, 2011). In this paper, we present a theoretical analysis on the approximation error and learning ability of the lp norm MKL. Our analysis shows explicit learning rates for lp-norm MKL and demonstrates some notable advantages compared with traditional kernel-based learning algorithms where the kernel is fixed. [ABSTRACT FROM AUTHOR]

In the era of big data, many urgent issues to tackle in all walks of life all can be solved via big data technique. Compared with the Internet, economy, industry, and aerospace fields, the application of big data in the area of architecture is relatively few. In this paper, on the basis of the actual data, the values of Boston suburb houses are forecast by several machine learning methods. According to the predictions, the government and developers can make decisions about whether developing the real estate on corresponding regions or not. In this paper, support vector machine (SVM), least squares support vector machine (LSSVM), and partial least squares (PLS)methods are used to forecast the home values. And these algorithms are compared according to the predicted results. Experiment shows that although the data set exists serious nonlinearity, the experiment result also show SVM and LSSVM methods are superior to PLS on dealing with the problem of nonlinearity. The global optimal solution can be found and best forecasting effect can be achieved by SVM because of solving a quadratic programming problem. In this paper, the different computation deficiencies of the algorithms are compared according to the computing times of relevant algorithms. [ABSTRACT FROM AUTHOR]

Chaotic Neural Network, also denoted by the acronym CNN, has rich dynamical behaviors that can be harnessed in promising engineering applications. However, due to its complex synapse learning rules and network structure, it is difficult to update its synaptic weights quickly and implement its large scale physical circuit. This paper addresses an implementation scheme of a novel CNN with memristive neural synapses that may provide a feasible solution for further development of CNN. Memristor, widely known as the fourth fundamental circuit element, was theoretically predicted by Chua in 1971 and has been developed in 2008 by the researchers in Hewlett-Packard Laboratory. Memristor based hybrid nanoscale CMOS technology is expected to revolutionize the digital and neuromorphic computation. The proposed memristive CNN has four significant features: (1) nanoscale memristors can simplify the synaptic circuit greatly and enable the synaptic weights update easily; (2) it can separate stored patterns from superimposed input; (3) it can deal with one-to-many associative memory; (4) it can deal with many-to-many associative memory. Simulation results are provided to illustrate the effectiveness of the proposed scheme. [ABSTRACT FROM AUTHOR]

Splitting methods have recently received much attention due to the fact that many nonlinear problems arising in applied areas such as image recovery, signal processing, and machine learning are mathematically modeled as a nonlinear operator equation and this operator is decomposed as the sum of two (possibly simpler) nonlinear operators. Most of the investigation on splitting methods is however carried out in the framework of Hilbert spaces. In this paper, we consider these methods in the setting of Banach spaces. We shall introduce two iterative forward-backward splitting methods with relaxations and errors to find zeros of the sum of two accretive operators in the Banach spaces. We shall prove the weak and strong convergence of these methods under mild conditions. We also discuss applications of these methods to variational inequalities, the split feasibility problem, and a constrained convex minimization problem. [ABSTRACT FROM AUTHOR]

Large data sets classification is widely used in many industrial applications. It is a challenging task to classify large data sets efficiently, accurately, and robustly, as large data sets always contain numerous instances with high dimensional feature space. In order to deal with this problem, in this paper we present an online Logdet divergence based metric learning (LDML) model by making use of the powerfulness of metric learning. We firstly generate a Mahalanobis matrix via learning the training data with LDML model. Meanwhile, we propose a compressed representation for high dimensional Mahalanobis matrix to reduce the computation complexity in each iteration. The final Mahalanobis matrix obtained this way measures the distances between instances accurately and serves as the basis of classifiers, for example, the k-nearest neighbors classifier. Experiments on benchmark data sets demonstrate that the proposed algorithm compares favorably with the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

In the ranking problem, one has to compare two different observations and decide the ordering between them. It has received increasing attention both in the statistical and machine learning literature. This paper considers l¹-regularized ranking rules with convex loss. Under some mild conditions, a learning rate is established. [ABSTRACT FROM AUTHOR]

The study of multitask learning algorithms is one of very important issues. This paper proposes a least-square regularized regression algorithm for multi-task learning with hypothesis space being the union of a sequence of Hilbert spaces. The algorithm consists of two steps of selecting the optimal Hilbert space and searching for the optimal function. We assume that the distributions of different tasks are related to a set of transformations under which any Hilbert space in the hypothesis space is norm invariant. We prove that under the above assumption the optimal prediction function of every task is in the same Hilbert space. Based on this result, a pivotal error decomposition is founded, which can use samples of related tasks to bound excess error of the target task. We obtain an upper bound for the sample error of related tasks, and based on this bound, potential faster learning rates are obtained compared to single-task learning algorithms. [ABSTRACT FROM AUTHOR]

This paper considers the stability of high-order PID-type iterative learning control law for a class of nonlinear switched systems with state delays and arbitrary switched rules, which perform a given task repeatedly. The stability condition for the proposed high-order learning control law is first established, and then the stability is analyzed based on contraction mapping approach in the sense of λ norm. It is shown that the proposed iterative learning control law can guarantee the asymptotic convergence of the tracking error for the entire time interval through the iterative learning process. Two examples are given to illustrate the effectiveness of the proposed approach. [ABSTRACT FROM AUTHOR]

Based on sparse representation, a robust global and local classification algorithm for visual target tracking in uncertain environment was proposed in this paper. The global region of target and the position of target would be found, respectively by the proposed algorithm. Besides, overcompleted dictionary was obtained and updated by biased discriminate analysis with the divergence of positive and negative samples at current frame. And this over-completed dictionary not only discriminates the positive samples accurately but also rejects the negative samples effectively. Experiments on challenging sequences with evaluation of the state-of the-art methods show that the proposed algorithm has better robustness to illumination changes, perspective changes, and targets rotation itself. [ABSTRACT FROM AUTHOR]

This paper proposed a crime busting model with two dynamic ranking algorithms to detect the likelihood of a suspect and the possibility of a leader in a complex social network. Signally, in order to obtain the priority list of suspects, an advanced network mining approach with a dynamic cumulative nominating algorithm is adopted to rapidly reduce computational expensiveness than most other topology-based approaches. Our method can also greatly increase the accuracy of solution with the enhancement of semantic learning filtering at the same time. Moreover, another dynamic algorithm of node contraction is also presented to help identify the leader among conspirators. Test results are given to verify the theoretical results, which show the great performance for either small or large datasets. [ABSTRACT FROM AUTHOR]

This paper considers the ERM scheme for quantile regression. We conduct error analysis for this learning algorithm by means of a variance-expectation bound when a noise condition is satisfied for the underlying probability measure. The learning rates are derived by applying concentration techniques involving the l²-empirical covering numbers. [ABSTRACT FROM AUTHOR]