Synthesis of systems of neurofunctional transformations in classification problems

Optimal synthesis of linear and nonlinear transformations is used to synthesize pattern recognition systems. The necessary and sufficient conditions for the existence of robust dichotomous linear separability of sets in feature space are represented in terms of pseudoinverse operations. The synthesis of classification systems is reduced to searching for the best nonlinear transformations of the components of the feature vector or optimal linear combinations of its components. __________ Translated from Kibernetika i Sistemnyi Analiz, No. 3, pp. 47–56, May–June 2007.     

Analysis and synthesis of signal classification systems by perturbing pseudoinverse and projection operations

The paper considers the synthesis of signal classification systems using the theory of perturbation of pseudoreverse and projection matrices and proposes an algorithm to select (filter) separable subsets on a finite point set from R ⁿ . Translated from Kibernetika i Sistemnyi Analiz, No. 4, pp. 124–134, July–August 2009.     

Development of expert control systems: A pattern classification and recognition approach

In this paper, a pattern classification and recognition approach to expert control systems is developed for use in the on-line analysis and design of dynamic systems. The approach used is based on the tuning of a three-term PID controller and, hence, it is not dependent on a specific form of the process model. A real-time experiment of implementing the developed controller using a microcomputer and associated hardware is presented. A sample set of production rules is discussed. The expert system reaches appropriate tuning parameters, using extracted features, such as oscillatory, underdamped, and exponentially monotonic properties.     

基于Adaboost和CART结合的优化分类算法

提出了一种基于Adaboost算法和CART算法结合的分类算法。以特征为节点生成CART二叉树,用CART二叉树代替传统Adaboost算法中的弱分类器,再由这些弱分类器生成强分类器。将强分类器对数字样本和人脸样本分类,与传统Adaboost算法相比,该方法的错误率分别减少20%和86.5%。将分类器应用于目标检测上,实现了对这两种目标的快速检测和定位。结果表明,改进算法既减小了对样本分类的错误率,又保持了传统Adboost算法对目标检测的快速性。     

基于流形距离的人工免疫无监督分类与识别算法

将一种新的流形距离作为相似性度量测度, 提出了一种用于无监督分类与识别的人工免疫系统方法. 通过基于流形距离的相似性度量, 有效利用样本集固有的全局一致性信息, 充分挖掘无类属样本的空间分布信息, 对样本进行类别划分. 新方法将免疫响应过程建模为一个四元组 AIR=(G,I,R,A) , 其中 G 为引发免疫响应的外界刺激, 即抗原; I 为所有可能抗体的集合; R 为抗体间相互作用的规则集合; A 为支配抗体反应、指导抗体进化的动态算法. 针对无监督分类问题, 将抗体编码为代表各类别的典型样本序号的排列, 利用动态算法 A 搜索能代表各类别的典型样本的最佳组合. 将新方法与标准的 K-均值算法、基于流形距离的进化聚类算法以及 Maulik 等人提出的基于遗传算法的聚类算法进行了性能比较. 对 6 个人工数据集及手写体数字识别问题的仿真实验结果显示, 新方法对样本空间分布复杂的无监督分类问题和实际的模式识别问题具有较高的准确率和较好的鲁棒性.     

Architectures for detecting and solving conflicts: two-stage classification and support vector classifiers

In the majority of cases, a properly trained classifier or ensemble of classifiers may yield acceptable recognition results. However, in some cases, recognition will fail due to typical conflicts that are encountered, like the confusion between [A] and [H] or [U] and [V]. In this paper, two architectures for the recognition of handwritten text are described. The key issue for each of these systems is to detect the event of a possible conflict and subsequently attempt to solve that particular problem. Both systems exploit a two-stage classification method. In the event that the first-stage classifiers are not certain about the result, the second-stage system engages a set of support vector classifiers for refining the output hypothesis.Received: 31 October 2001, Accepted: 13 December 2002, Published online: 6 June 2003     

电子鼻传感器阵列优化及其在醋的分类中的应用

利用6个TGS系列气敏传感器阵列对醋进行了检测。按照Wilks统计量最小的原则对传感器阵列进行了优化,得到了由4个传感器组成的用于检测醋的种类的最佳传感器阵列。对阵列优化前后的数据,用PCA、LDA进行对比研究,结果表明,优化后的阵列可以更好地对醋鉴别分类。因此,所给出的优化方法是有效的。     

Peptide classification using optimal and information theoretic syntactic modeling

We consider the problem of classifying peptides using the information residing in their syntactic representations. This problem, which has been studied for more than a decade, has typically been investigated using distance-based metrics that involve the edit operations required in the peptide comparisons. In this paper, we shall demonstrate that the Optimal and Information Theoretic (OIT) model of Oommen and Kashyap [22] applicable for syntactic pattern recognition can be used to tackle peptide classification problem. We advocate that one can model the differences between compared strings as a mutation model consisting of random substitutions, insertions and deletions obeying the OIT model. Thus, in this paper, we show that the probability measure obtained from the OIT model can be perceived as a sequence similarity metric, using which a support vector machine (SVM)-based peptide classifier can be devised. The classifier, which we have built has been tested for eight different substitution matrices and for two different data sets, namely, the HIV-1 Protease cleavage sites and the T-cell epitopes. The results show that the OIT model performs significantly better than the one which uses a Needleman-Wunsch sequence alignment score, it is less sensitive to the substitution matrix than the other methods compared, and that when combined with a SVM, is among the best peptide classification methods available.     

Generative learning of visual concepts using multiobjective genetic programming

This paper introduces a novel method of visual learning based on genetic programming, which evolves a population of individuals (image analysis programs) that process attributed visual primitives derived from raw raster images. The goal is to evolve an image analysis program that correctly recognizes the training concept (shape). The approach uses generative evaluation scheme: individuals are rewarded for reproducing the shape of the object being recognized using graphical primitives and elementary background knowledge encoded in predefined operators. Evolutionary run is driven by a multiobjective fitness function to prevent premature convergence and enable effective exploration of the space of solutions. We present the method in detail and verify it experimentally on the task of learning two visual concepts from examples.     

基于原型与正交投影学习的图像集分类算法

为了利用图像集中的集合信息来提高图像识别精度以及对图像变化的鲁棒性,从而大幅降低诸如姿态、光照、遮挡和未对齐等因素对识别精度的影响,提出了一种用于图像集分类的图像集原型与投影学习算法（LPSOP）。该算法针对每个图像集学习有代表性的点（原型）以及一个正交的全局投影矩阵,使得在目标子空间的每个图像集可以被最优地分类到同类的最近原型集中。用学习到的原型来代表该图像集,既能降低冗余图像干扰,又能减少存储和计算开销,学习到的投影矩阵则能够大幅提高分类精度与噪声鲁棒性。在UCSD/Honda、CMU MoBo和YouTube celebrities这三个数据集上的实验结果表明,LPSOP比目前流行的图像集分类算法具有更高的识别精度和更好的鲁棒性。     

Pseudoinverse and Projection Matrices in Problems of Synthesis of Functional Transformers

An approach is proposed to derive dependences in the form of functional transformers. Conditions of existence and the method of constructing such functional transformers as a superposition of linear and nonlinear functions from a specified class are presented. The principle of optimal synthesis of functional transformers is formulated based on the theory of disturbance of projective and pseudoinverse operators. Methods are proposed for constructing complex functional transformers of different topologies.     

一种反馈过程神经元网络模型及在动态信号分类中的应用*

针对动态信号模式分类问题,提出了一种反馈过程神经元网络模型和基于该模型的分类方法。这种网络的输入可直接为时变函数,网络的信息传输既有与前馈神经元网络一样的前向流,也有后面各层节点到前层节点的反馈,且可对节点自身反馈输出信息,能直接用于动态信号的模式分类。由于反馈过程神经元网络在对输入样本的学习中增加了神经元输出信息的反馈,可提高网络的学习效率和稳定性。给出了具体学习算法,以时变函数样本集的分类问题为例,实验结果验证了模型和算法的有效性。     

Bayes分类技术在计算机病毒检测中的研究

本文从恶意程序检测的问题入手,介绍了病毒检测技术中的问题和难点,通过对恶意程序的主要特征分析,结合当前迅速发展的模式识别与智能检测成果,提出了使用Bayes分类法对恶意程序进行判断的方案,并对该方案的优缺点进行了分析。     

Determining the characteristics of systems with multiple inputs and multiple outputs

A simple and computationally efficient method of determining the transfer function and other characteristics of systems with multiple inputs and multiple outputs (MIMO systems) described by autoregression equations is proposed. The method is based on the use of the discrete Fourier transformation. The algorithm is highly suitable for computer implementation. The efficiency and simplicity of the method are illustrated using the example of a system with three inputs and three outputs. The proposed concept may be applied to systems described by the autoregression sliding mean.     

Modal synthesis of multivariable systems with observers of specified order

The problem of pole assignment in multivariable linear systems with observers of a given order v is considered. By way of a simple proof, a very general result is established which constitutes an extension of all theorems known in the literature on pole assignment via observers. On its basis, an efficient design algorithm is worked out.     

Improving the performance of fuzzy rule-based classification systems with interval-valued fuzzy sets and genetic amplitude tuning

Among the computational intelligence techniques employed to solve classification problems, Fuzzy Rule-Based Classification Systems (FRBCSs) are a popular tool because of their interpretable models based on linguistic variables, which are easier to understand for the experts or end-users.The aim of this paper is to enhance the performance of FRBCSs by extending the Knowledge Base with the application of the concept of Interval-Valued Fuzzy Sets (IVFSs). We consider a post-processing genetic tuning step that adjusts the amplitude of the upper bound of the IVFS to contextualize the fuzzy partitions and to obtain a most accurate solution to the problem.We analyze the goodness of this approach using two basic and well-known fuzzy rule learning algorithms, the Chi et al.’s method and the fuzzy hybrid genetics-based machine learning algorithm. We show the improvement achieved by this model through an extensive empirical study with a large collection of data-sets.     

Robust analysis and synthesis for a class of uncertain nonlinear systems with multiple equilibria

This paper focuses on a class of uncertain nonlinear systems which are subject to norm-bounded parameter uncertainty in the forward path and a vector-valued periodic nonlinearity in the feedback path, and addresses robust analysis and synthesis problems for such systems. Sufficient conditions for global asymptotic stability are derived in terms of linear matrix inequalities (LMIs) and a technique for the estimation of the uncertainty bound is proposed by solving a generalized eigenvalue minimization problem. The problem of robust synthesis is concerned with designing a feedback controller such that the resulting closed-loop system is globally asymptotically stable for all admissible uncertainties. It is shown that a solution to the robust synthesis problem for the uncertain system can be obtained by solving a synthesis problem for an uncertainty free system. A concrete example is presented to demonstrate the applicability and validity of the proposed approach.     

多维有序聚类法在地质数据分类中的应用

针对经验法在地质多指标数据进行有序分类存在局限性的问题,提出了基于Fisher最优分割的多维有序聚类法,通过对均一化处理的多维指标之间定义类直径的损失函数来判别最优分割,然后在使得损失函数最小的情况下合并相邻分类,实现有序样本数据的聚类层次。采用该方法对钻孔编录地层划分、岩体质量分级和孢粉粒度分带三种数据进行分类,结果表明在不能确定分层数的情况下,可以利用损失函数的导数曲线极值点确定最优分层数;样品多个指标之间的相关性越高,分类结果与单指标分类结果的差异越不明显。该方法能够为多元地质数据的定量分析提供数学理论依据,也可以应用于其他行业领域。     

Concept of structure-and-objective analysis and synthesis of organization-and-engineering systems

A concept of structure-and-objective analysis and synthesis of organization-and-engineering industrial systems is defined. The logical aspect of system paradigm and principles of structure-and-objective analysis is discussed. It allows us to resolve the problem of disagreement of results of systems analysis. __________ Translated from Kibernetika i Sistemnyi Analiz, No. 6, pp. 147–156, November–December 2006.     

Structure-based object representation and classification in mobile robotics through a Microsoft Kinect

A new approach enabling a mobile robot to recognize and classify furniture-like objects composed of assembled parts using a Microsoft Kinect is presented. Starting from considerations about the structure of furniture-like objects, i.e., objects which can play a role in the course of a mobile robot mission, the 3D point cloud returned by the Kinect is first segmented into a set of “almost convex” clusters. Objects are then represented by means of a graph expressing mutual relationships between such clusters. Off-line, snapshots of the same object taken from different positions are processed and merged, in order to produce multiple-view models that are used to populate a database. On-line, as soon as a new object is observed, a run-time window of subsequent snapshots is used to search for a correspondence in the database.Experiments validating the approach with a set of objects (i.e., chairs, tables, but also other robots) are reported and discussed in detail.