一种基于高阶累积量的因果结构学习算法

从观测数据中学习因果结构具有重要的应用价值。目前,一类学习因果结构的方法是基于函数因果模型假设,通过检验噪声与原因变量的独立性来学习因果结构。然而,该类方法涉及高计算复杂度的独立性检验过程,影响结构学习算法的实用性和鲁棒性。为此,提出了一种在线性非高斯模型下,利用高阶累积量作为独立性评估的因果结构学习算法。该算法主要分为两个步骤,第一个步骤是利用基于条件独立性约束的方法学习到因果结构的马尔可夫等价类,第二个步骤是定义了一种基于高阶累积量的得分,该得分可以判别两个随机变量的独立性,从而可以从马尔可夫等价类中搜索到最佳独立性得分的因果结构作为算法的输出。该算法的优势在于：a)相比基于核方法的独立性检验,该方法有较低的计算复杂度;b)基于得分搜索的方法,可以得到一个最匹配数据生成过程的模型,提高学习方法的鲁棒性。实验结果表明,基于高阶累积量的因果结构学习方法在合成数据中F₁得分提高了5%,并在真实数据中学习到更多的因果方向。     

基于因果语义定向的贝叶斯网络结构学习

基于变量之间基本依赖关系、基本结构、d-separation标准、依赖分析思想和混合定向策略,给出了一种有效实用的贝叶斯网络结构学习方法,不需要结点有序,并能避免打分-搜索方法存在的指数复杂性,以及现有依赖分析方法的大量高维条件概率计算等问题。     

A variational integrators approach to second order modeling and identification of linear mechanical systems

The theory of variational integration provides a systematic procedure to discretize the equations of motion of a mechanical system, preserving key properties of the continuous time flow. The discrete-time model obtained by variational integration theory inherits structural conditions which in general are not guaranteed under general discretization procedures. We discuss a simple class of variational integrators for linear second order mechanical systems and propose a constrained identification technique which employs simple linear transformation formulas to recover the continuous time parameters of the system from the discrete-time identified model. We test this approach on a simulated eight degrees of freedom system and show that the new procedure leads to an accurate identification of the continuous-time parameters of second-order mechanical systems starting from discrete measured data.     

A new kernel-based approach for linear system identification

This paper describes a new kernel-based approach for linear system identification of stable systems. We model the impulse response as the realization of a Gaussian process whose statistics, differently from previously adopted priors, include information not only on smoothness but also on BIBO-stability. The associated autocovariance defines what we call a stable spline kernel. The corresponding minimum variance estimate belongs to a reproducing kernel Hilbert space which is spectrally characterized. Compared to parametric identification techniques, the impulse response of the system is searched for within an infinite-dimensional space, dense in the space of continuous functions. Overparametrization is avoided by tuning few hyperparameters via marginal likelihood maximization. The proposed approach may prove particularly useful in the context of robust identification in order to obtain reduced order models by exploiting a two-step procedure that projects the nonparametric estimate onto the space of nominal models. The continuous-time derivation immediately extends to the discrete-time case. On several continuous- and discrete-time benchmarks taken from the literature the proposed approach compares very favorably with the existing parametric and nonparametric techniques.     

A new generator of causal ideal internal dynamics for a class of unstable linear differential equations

The generator design for causal ideal internal dynamics (IID), namely, solving IID, is a fundamental problem in a nonminimum‐phase output tracking process. In this paper, for a class of unstable matrix differential equations, a new causal IID generator is proposed, whose parameters are partly chosen via optimization. Compared with existing similar design schemes, it is applicable to matrix differential equations with singular system matrices. Also, it requires less computation, avoids taking higher order derivatives, and can be easily extended to treat slowly time‐varying matrix differential equations without the need for extra computation. Copyright © 2016 John Wiley & Sons, Ltd.     

A new approach to model virtual channels in interconnection networks

Dealing with virtual channels has always been a critical issue in developing analytical performance models for interconnection networks. Almost all previous studies relied on a method proposed by Dally to capture the effect of virtual channels multiplexing in the performance of interconnection networks. This paper presents a new method to model the effect of virtual channel multiplexing in high-speed wormhole-switched interconnection networks. Dally's method loses its accuracy as the traffic load increases due to blocking nature of wormhole-switched networks. Our new method is based on a finite capacity queue, M/G/1/V and comparing to Dally's method achieves a higher degree of accuracy under low, moderate and high traffic loads. Furthermore, its simplicity eases its employment under different network conditions and setup. The presented model is validated by means of an event driven simulator and a detailed comparison with Dally's method is presented.     

一种可度量的贝叶斯网络结构学习方法

针对基于约束的方法存在的序依赖、高阶检验等问题,提出了一种通过互信息排序的贝叶斯网络结构学习方法,该方法包括度量信息矩阵学习和“偷懒”启发式策略2部分.其中度量信息矩阵刻画了变量间的依赖程度而且暗含了程度强弱的比较,有效地解决了检验过程中由于变量序导致的误判问题;“偷懒”启发式策略在度量信息矩阵的指导下有选择地将变量加入到条件集中,有效地降低了高阶检验而且减少了检验次数.从理论上证明了新方法的可靠性,从实验上展示了在不丢失学习结构质量的条件下,新方法的搜索比其他搜索过程显著快而且易扩展到样本量小且稀疏的数据集上.     

HiWalk: Learning node embeddings from heterogeneous networks

Heterogeneous networks, such as bibliographical networks and online business networks, are ubiquitous in everyday life. Nevertheless, analyzing them for high-level semantic understanding still poses a great challenge for modern information systems. In this paper, we propose HiWalk to learn distributed vector representations of the nodes in heterogeneous networks. HiWalk is inspired by the state-of-the-art representation learning algorithms employed in the context of both homogeneous networks and heterogeneous networks, based on word embedding learning models. Different from existing methods in the literature, the purpose of HiWalk is to learn vector representations of the targeted set of nodes by leveraging the other nodes as “background knowledge”, which maximizes the structural correlations of contiguous nodes. HiWalk decomposes the adjacent probabilities of the nodes and adopts a hierarchical random walk strategy, which makes it more effective, efficient and concentrated when applied to practical large-scale heterogeneous networks. HiWalk can be widely applied in heterogeneous networks environments to analyze targeted types of nodes. We further validate the effectiveness of the proposed HiWalk through multiple tasks conducted on two real-world datasets.     

Learning with personalized recommender systems: A psychological view

This paper explores the potentials of recommender systems for learning from a psychological point of view. It is argued that main features of recommender systems (collective responsibility, collective intelligence, user control, guidance, personalization) fit very well to principles in the learning sciences. However, recommender systems should not be transferred from commercial to educational contexts on a one-to-one basis, but rather need adaptations in order to facilitate learning. Potential adaptations are discussed both with regard to learners as recipients of information and learners as producers of data. Moreover, it is distinguished between system-centered adaptations that enable proper functioning in educational contexts, and social adaptations that address typical information processing biases. Implications for the design of educational recommender systems and for research on educational recommender systems are discussed.     

A methodology for control-relevant nonlinear system identification using restricted complexity models

A broadly-applicable, control-relevant system identification methodology for nonlinear restricted complexity models (RCMs) is presented. Control design based on RCMs often leads to controllers which are easy to interpret and implement in real-time. A control-relevant identification method is developed to minimize the degradation in closed-loop performance as a result of RCM approximation error. A two-stage identification procedure is presented. First, a nonlinear ARX model is estimated from plant data using an orthogonal least squares algorithm; a Volterra series model is then generated from the nonlinear ARX model. In the second stage, a RCM with the desired structure is estimated from the Volterra series model through a model reduction algorithm that takes into account closed-loop performance requirements. The effectiveness of the proposed method is illustrated using two chemical reactor examples.     

A new identification method for fuzzy linear models of nonlinear dynamic systems

The most promising methods for identifying a fuzzy model are data clustering, cluster merging and subsequent projection of the clusters on the input variable space. This article proposes to modify this procedure by adding a cluster rotation step, and a method for the direct calculation of the consequence parameters of the fuzzy linear model. These two additional steps make the model identification procedure more accurate and limits the loss of information during the identification procedure. The proposed method has been tested on a nonlinear first order model and a nonlinear model of a bioreactor and results are very promising.     

A new structure adaptation algorithm for RBF networks and its application

An adaptation algorithm is developed for radial basis function network (RBFN) in this paper. The RBFN is adapted on-line for both model structure and parameters with measurement data. When the RBFN is used to model a non-linear dynamic system, the structure is adapted to model abrupt change of system operating region, while the weights are adapted to model the incipient time varying parameters. Two new algorithms are proposed for adding new centres while the redundant centres are pruned, which is particularly useful for model-based control. The developed algorithm is evaluated by modelling a numerical example and a chemical reactor rig. The performance is compared with a non-adaptive model.     

set membership identification: A survey

Robustness had become in past years a central issue in system and control theory, focusing the attention of researchers from the study of a single model to the investigation of a set of models, described by a set of perturbations of a “nominal” model. Such a set, often indicated as an uncertainty model set or model set for short, has to be suitably constructed to describe the inherent uncertainty about the system under consideration and to be used for analysis and design purposes. H_∞ identification methods deliver uncertainty model sets in a suitable form to be used by well-established robust design techniques, based on H_∞ or μ optimization methods. The literature on H_∞ identification is now very extensive. In this paper, some of the most relevant contributions related to assumption validation, evaluation of bounds on unmodeled dynamics, convergence analysis and optimality properties of linear, two-stage and interpolatory algorithms are surveyed from a deterministic point of view.     

A new technique to sense non-Gaussian features of the sea surface from L-band bi-static GNSS reflections

We present a new technique to analyze L-band signals of the Global Navigation Satellite System (GNSS) that have rebounded off the sea surface, with the aim of retrieving information about surface roughness in the form of the effective Probability Density Function (PDF) of the slopes. Unlike earlier techniques, which parameterize the PDF (usually as normal bivariate distributions), this approach does not constrain the surface slopes' PDF to the shape of a particular analytical distribution. This may help to understand the real information content of L-band scattered signals, which is currently unclear. After validating the algorithm by means of end-to-end simulations, we have applied it to real data. The tests on real data show that the retrievals are robust, consistent with the results obtained with standard GNSS-reflection techniques, and in agreement with independent sources of information. Moreover, the retrieved slopes' PDF present non-Gaussian features, such as skewness. A more in-depth analysis to check whether the skewness is a geophysical signature or an artifact of the technique, shows that it maps with the up-/down-asymmetries introduced by surface forces. In particular, this is the first time that GNSS-reflections have sensed and identified the up- and down-wind signature on the surface, putting and end to the 180° ambiguity that was usually attributed to GNSS observations of directional roughness.     

A novel hybrid classification model of artificial neural networks and multiple linear regression models

The classification problem of assigning several observations into different disjoint groups plays an important role in business decision making and many other areas. Developing more accurate and widely applicable classification models has significant implications in these areas. It is the reason that despite of the numerous classification models available, the research for improving the effectiveness of these models has never stopped. Combining several models or using hybrid models has become a common practice in order to overcome the deficiencies of single models and can be an effective way of improving upon their predictive performance, especially when the models in combination are quite different. In this paper, a novel hybridization of artificial neural networks (ANNs) is proposed using multiple linear regression models in order to yield more general and more accurate model than traditional artificial neural networks for solving classification problems. Empirical results indicate that the proposed hybrid model exhibits effectively improved classification accuracy in comparison with traditional artificial neural networks and also some other classification models such as linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), K-nearest neighbor (KNN), and support vector machines (SVMs) using benchmark and real-world application data sets. These data sets vary in the number of classes (two versus multiple) and the source of the data (synthetic versus real-world). Therefore, it can be applied as an appropriate alternate approach for solving classification problems, specifically when higher forecasting accuracy is needed.     

A review of the Expectation Maximization algorithm in data-driven process identification

The Expectation Maximization (EM) algorithm has been widely used for parameter estimation in data-driven process identification. EM is an algorithm for maximum likelihood estimation of parameters and ensures convergence of the likelihood function. In presence of missing variables and in ill conditioned problems, EM algorithm greatly assists the design of more robust identification algorithms. Such situations frequently occur in industrial environments. Missing observations due to sensor malfunctions, multiple process operating conditions and unknown time delay information are some of the examples that can resort to the EM algorithm. In this article, a review on applications of the EM algorithm to address such issues is provided. Future applications of EM algorithm as well as some open problems are also provided.     

Construction and implementation of general linear methods for ordinary differential equations: A review

It it the purpose of this paper to review the results on the construction and implementation of diagonally implicit multistage integration methods for ordinary differential equations. The systematic approach to the construction of these methods with Runge-Kutta stability is described. The estimation of local discretization error for both explicit and implicit methods is discussed. The other implementations issues such as the construction of continuous extensions, stepsize and order changing strategy, and solving the systems of nonlinear equations which arise in implicit schemes are also addressed. The performance of experimental codes based on these methods is briefly discussed and compared with codes from Matlab ordinary differential equation (ODE) suite. The recent work on general linear methods with inherent Runge-Kutta stability is also briefly discussed.     

A new view of anti‐windup design for uncertain linear systems in the frequency domain

This paper presents a new perspective on the stability problem for uncertain LTI feedback systems with actuator input amplitude saturation. The solution is obtained using the quantitative feedback theory and a 3 DoF non‐interfering control structure. Describing function (DF) analysis is used as a criterion for closed loop stability and limit cycle avoidance, but the circle or Popov criteria could also be employed. The novelty is the combination of a controller parameterization from the literature and describing function‐based limit cycle avoidance with margins for uncertain plants. Two examples are given. The first is a benchmark problem and a comparison is made with other proposed solutions. The second is an example that was implemented and tested on an X‐Y linear stage used for nano‐positioning applications. Design and implementation considerations are given. An example is given on how the method can be extended to amplitude and rate saturation with the help of the generalized describing function, and a novel anti‐windup compensation structure inspired by previous contributions. Copyright © 2015 John Wiley & Sons, Ltd.     

Information retrieval: a view from the Chinese IR community

During a two day strategic workshop in February 2018,22 information retrieval researchers met to discuss the future challenges and opportunities within the field.The outcome is a list of potential research directions,project ideas,and challenges.This report describes the major conclusions we have obtained during the workshop.A key result is that we need to open our mind to embrace a broader IR field by rethink the definition of information,retrieval,user,system,and evaluation of IR.By providing detailed discussions on these topics,this report is expected to inspire our IR researchers in both academia and industry,and help the future growth of the IR research community.