Mapping graphs Of parallel programs onto graphs of distributed computer systems by recurrent neural networks

A problem of mapping graphs of parallel programs onto graphs of distributed computer systems by recurrent neural network is formulated. Parameter values providing absence of incorrect solutions are experimentally determined. Because of introduction of penalty coefficient into Lyapunov function for the program graph edges non-coincided with the system graph edges, optimal solutions are found for mapping a “line”-graph onto a two-dimensional torus. For increasing probability of finding optimal mapping, a method for splitting the mapping is proposed. The method essence is a reducing solution matrix to a block-diagonal form. The Wang recurrent neural network is used to exclude incorrect solutions of the problem of mapping the line-graph onto a three-dimensional torus. This network converges quicker than the Hopfield one.     

基于神经网络的纹理图象生成

提出了一种利用多层前馈神经网络生成纹理图象的新方法 .利用该方法可方便地生成图案丰富的纹理图象集 ,并且该纹理图象集中的任何一幅图象均唯一地对应一组神经网络的权值和阈值 ,因此不仅便于图象保存 ,还大大地节省了图象存储空间 .     

Learning from interpretation transition

We propose a novel framework for learning normal logic programs from transitions of interpretations. Given a set of pairs of interpretations (I,J) such that J=T _P (I), where T _P is the immediate consequence operator, we infer the program P. The learning framework can be repeatedly applied for identifying Boolean networks from basins of attraction. Two algorithms have been implemented for this learning task, and are compared using examples from the biological literature. We also show how to incorporate background knowledge and inductive biases, then apply the framework to learning transition rules of cellular automata.     

A hybrid radial basis function and data envelopment analysis neural network for classification

We propose a hybrid radial basis function network-data envelopment analysis (RBFN-DEA) neural network for classification problems. The procedure uses the radial basis function to map low dimensional input data from input space ℜ to a high dimensional ℜ⁺ feature space where DEA can be used to learn the classification function. Using simulated datasets for a non-linearly separable binary classification problem, we illustrate how the RBFN-DEA neural network can be used to solve it. We also show how asymmetric misclassification costs can be incorporated in the hybrid RBFN-DEA model. Our preliminary experiments comparing the RBFN-DEA with feed forward and probabilistic neural networks show that the RBFN-DEA fares very well.     

Multistability of periodic delayed recurrent neural network with memristors

This paper discusses the recurrent neural network (RNN) with memristors as connection weights. Memristor is a nonlinear resistor. Memristance varies periodically with time when the sinusoidal voltage source is applied. According to this property of memristor, it shows that coefficients of RNN with memristors are periodic functions with respect to time t. By dividing the state space and using contraction mapping theorem, one sufficient condition is obtained for multiperiodicity. And the periodic orbits located in saturation regions are locally exponentially stable limit cycles. At last, one example is given for verifying the validity of our result.     

Neural network solution for finite-horizon H ∞ state feedback control of nonlinear systems

In this article, neural networks are used to approximately solve the finite-horizon optimal H _∞ state feedback control problem. The method is based on solving a related Hamilton–Jacobi–Isaacs equation of the corresponding finite-horizon zero-sum game. The neural network approximates the corresponding game value function on a certain domain of the state-space and results in a control computed as the output of a neural network. It is shown that the neural network approximation converges uniformly to the game-value function and the resulting controller provides closed-loop stability and bounded L ₂ gain. The result is a nearly exact H _∞ feedback controller with time-varying coefficients that is solved a priori offline. The results of this article are applied to the rotational/translational actuator benchmark nonlinear control problem.     

An adiabatic neural network for RBF approximation

Numerous studies have addressed nonlinear functional approximation by multilayer perceptrons (MLPs) and RBF networks as a special case of the more general mapping problem. The performance of both these supervised network models intimately depends on the efficiency of their learning process. This paper presents an unsupervised recurrent neural network, based on the recurrent Mean Field Theory (MFT) network model, that finds a least-squares approximation to an arbitrary L₂ function, given a set of Gaussian radially symmetric basis functions (RBFs). Essential is the reformulation of RBF approximation as a problem of constrained optimisation. A new concept of adiabatic network organisation is introduced. Together with an adaptive mechanism of temperature control this allows the network to build a hierarchical multiresolution approximation with preservation of the global optimisation characteristics. A revised problem mapping results in a position invariant local interconnectivity pattern, which makes the network attractive for electronic implementation. The dynamics and performance of the network are illustrated by numerical simulation.     

Incremental supervised learning for mobile robot reactive control

Reactive control for a mobile robot can be defined as a mapping from a perceptual space to a command space. This mapping can be hard-coded by the user (potential fields, fuzzy logic), and can also be learnt. This paper is concerned with supervised learning for perception to action mapping for a mobile robot. Among the existing neural approaches for supervised learning of a function, we have selected the grow and learn network for its properties adapted to robotic problems: incrementality and flexible structure. We will present the results we have obtained with this network using first raw sensor data and then pre-processed measures with the automatic construction of virtual sensors.     

Complexities of Homomorphism and Isomorphism for Definite Logic Programs

A homomorphism (?) of logic programs from P to P' is a function mapping Atoms(P) to Atoms(P') and it preserves complements and program clauses. For each definite program clause a←a1,...,an∈P it implies that (?)(a)←(?)(a1),...,(?)(an) is a program clause of P'. A homomorphism (?) is an isomorphism if (?) is a bijection. In this paper, the complexity of the decision problems on homomorphism and isomorphism for definite logic programs is studied. It is shown that the homomorphism problem (HOM-LP) for definite logic programs is NP-complete, and the isomorphism problem (ISO-LP) is equivalent to the graph isomorphism problem (GI).     

采用遗传算法训练对角递归神经网络预测控制器

本文提出了一种基于广义预测控制的神经网络预测控制方案．预测控制器由对角递归 神经网络预测控制器和前向神经网络静态补偿器组成．两种神经网络均采用遗传算法进行训 练．仿真实验表明,对于带纯时延的非线性被控对象,采用遗传算法设计的对角递归神经网 络预测控制器具有令人满意的控制性能．     

Peer-to-Peer in Metric Space and Semantic Space

This paper first proposes three improved gossip mechanisms by mapping links into metric space and dynamically adapting the number of selected neighbors to disseminate messages. Experiments and comparisons show that these mechanisms can improve the performance of gossip in peer-to-peer (P2P) networks. This is the effect of mapping a network into a metric space that differentiates nodes and links according to linking characteristics and controlling local information flow with knowing such differences. A further study about query routing on P2P semantic link network shows that mapping a network into a semantic space can also improve the performance. An intrinsic rule is found by experimental comparisons and analysis: the performance of a P2P network can be improved by designing an appropriate mapping from the network into metric space or semantic space. A general framework for networking with metric space and semantic space is suggested     

基于反馈神经网络的动态化工过程建模

针对非线性动态化工过程建模存在的问题,提出了一种新的反馈神经网络结构,并将状态反馈、时间序列延尺以及集中节点的概念结合起来,用于提高反馈神经网络的性能,同时又使得网络结构不至于太复杂,在用此网络结构建模的时间,成功地将BP算法用一网络模型的训练。文中将这种反馈神经网络结构分别对一个单输入单输出（SISO）的非线笥动态系统和一个多输入单输出（SIMO）的连续全混釜（CSTR）模型进行建模,并将所得模型与基于表态BP神经神经所得的模型在模型输出精度和抗干扰性等方面进行了比较,证明了该反馈神经在动态过程建模中能够比静态BP模型更好地反映出动态过程的输入输出关系,并具有一定的抗干扰能力。     

基于改进深度度量学习算法的表面缺陷检测

为了解决小批量、多品种工业产品的表面质量检测问题,提出一种基于改进深度度量学习的缺陷检测算法。该算法对VGG16网络模型做改进,更有利于原始图像的隐空间映射。针对产品表面缺陷检测的任务,提出条件三元组损失函数以加强神经网络的拟合能力。同时,在隐空间中进行缺陷判定时,抛弃原始度量学习中基于KNN算法的归类方法,提出基于高斯分布概率的归类模型。在检测新类型产品时,在已训练好的网络模型的基础上,使用新产品的图像数据作为输入对网络进行微调。利用该算法在纽扣缺陷数据集上经K-Fold交叉验证,在只需50个无缺陷样本和50个有缺陷样本的小样本情况下,该算法在不同的查询集上的检测准确率均在90%以上,最高可达99.89%,与传统深度度量学习算法相比,检测准确率提升10%以上。实验结果表明,改进深度度量学习算法可以很好地解决小批量、多品种工业产品的表面缺陷检测问题。     

一种自适应预测非平稳信号的新方法

本文提出一种动态神经网络为非平稳信号作自适应单步预测，它由级联回归神经网络和抽头延时线组成．用它对非线性动态方程产生的时间序列作自适应预测，实验结果表明，其效果远远超过了传统的前馈神经网络．     

Prediction of reservoir sensitivity using RBF neural network with trainable radial basis function

Reservoir sensitivity prediction is an important basis for designing reservoir protection program scientifically and exploiting oil and gas resources efficiently. Researchers have long endeavored to establish a method to predict reservoir sensitivity, but all of the methods have some limitations. Radial basis function (RBF) neural network, which provided a powerful technique to model non-linear mapping and the learning algorithm for RBF neural networks, corresponds to the solution of a linear problem, therefore it is unnecessary to establish an accurate model or organize rules in large number, and it enjoys the advantages such as simple network structure, fast convergence rate, and strong approximation ability, etc. However, different radial basis function has different non-linear mapping ability, and different data require different radial basis functions. Nowadays, the choice of radial basis function in the network is based on experience or test result only, which exerts a great adverse impact on the network performance. In this study, a new RBF neural network with trainable radial basis function was proposed by the linear combination of common radial basis functions. The input parameters of the network were the influence factors of reservoir sensitivity such as porosity and permeability, etc. The output parameter was the corresponding sensitivity index. The network was trained and tested with the data collected from our own experiments. The results showed that the new RBF neural network is effective and improved, of which the accuracy is obviously higher than the network with single radial basis function for the prediction of reservoir sensitivity.     

Modelling and simulation of combinational digital circuits using Petri nets

The P-model approach of modeling a combinational digital network using Petri nets is introduced. In this model a given logic circuit its represented by a graph With only two types of nodes, places and transitions. A logic 1 value in any line of the circuit corresponds to the presence of a so-called token, and a logic 0 corresponds to the absence of tokens, in the corresponding place of the P-model. The operation of the circuit is reflected in the execution of the P-model resulting from the firing of transitions. Several minimizing transformations in the P-model domain are discussed, which reduces the number of places and transitions by an order of magnitude. Based on this P-model representation of a logic circuit, a logic simulation algorithm is outlined. The method is faster and also simpler to implement than conventional simulation techniques.     

Back propagation genetic and recurrent neural network applications in modelling and analysis of squeeze casting process

Today, in competitive manufacturing environment reducing casting defects with improved mechanical properties is of industrial relevance. This led the present work to deal with developing the input-output relationship in squeeze casting process utilizing the neural network based forward and reverse mapping. Forward mapping is aimed to predict the casting quality (such as density, hardness and secondary dendrite arm spacing) for the known combination of casting variables (that is, squeeze pressure, pressure duration, die and pouring temperature). Conversely, attempt is also made to determine the appropriate set of casting variables for the required casting quality (that is, reverse mapping). Forward and reverse mapping tasks are carried out utilizing back propagation, recurrent and genetic algorithm tuned neural networks. Parameter study has been conducted to adjust and optimize the neural network parameters utilizing the batch mode of training. Since, batch mode of training requires huge data, the training data is generated artificially using response equations. Furthermore, neural network prediction performances are compared among themselves (reverse mapping) and with those of statistical regression models (forward mapping) with the help of test cases. The results shown all developed neural network models in both forward and reverse mappings are capable of making effective predictions. The results obtained will help the foundry personnel to automate and précised control of squeeze casting process.     

The Connectionist Inductive Learning and Logic Programming System

This paper presents the Connectionist Inductive Learning and Logic Programming System (C-IL²P). C-IL²P is a new massively parallel computational model based on a feedforward Artificial Neural Network that integrates inductive learning from examples and background knowledge, with deductive learning from Logic Programming. Starting with the background knowledge represented by a propositional logic program, a translation algorithm is applied generating a neural network that can be trained with examples. The results obtained with this refined network can be explained by extracting a revised logic program from it. Moreover, the neural network computes the stable model of the logic program inserted in it as background knowledge, or learned with the examples, thus functioning as a parallel system for Logic Programming. We have successfully applied C-IL²P to two real-world problems of computational biology, specifically DNA sequence analyses. Comparisons with the results obtained by some of the main neural, symbolic, and hybrid inductive learning systems, using the same domain knowledge, show the effectiveness of C-IL²P.     

Back-propagation learning in expert networks

Expert networks are event-driven, acyclic networks of neural objects derived from expert systems. The neural objects process information through a nonlinear combining function that is different from, and more complex than, typical neural network node processors. The authors develop back-propagation learning for acyclic, event-driven networks in general and derive a specific algorithm for learning in EMYCIN-derived expert networks. The algorithm combines back-propagation learning with other features of expert networks, including calculation of gradients of the nonlinear combining functions and the hypercube nature of the knowledge space. It offers automation of the knowledge acquisition task for certainty factors, often the most difficult part of knowledge extraction. Results of testing the learning algorithm with a medium-scale (97-node) expert network are presented.     

Program partition and logic program analysis

A program partition scheme for stratified programs introduced by Apt et al. (1988) is used to study efficient computation of logic programs. We consider three types of program partitions and their corresponding graph representations: 1) the natural partition, 2) stratified partitions, and 3) the reduced partition. The natural (program) partition consists of definitions of relations, each definition being a subprogram. Subprograms of a program partition may consist of several relations. A partition graph is introduced for a program partition, each node of which corresponds to a subprogram. The partition graph for a stratified partition is a directed acyclic graph (DAG). A stratified partition decomposes a program into modules. The stratified partition with the maximum number of modules is the reduced partition. The cost to achieve a reduced partition is linear in the program size, using well known graph algorithms. We introduce the modular interpretations, which are equivalent in semantics to the standard interpretation. The modular interpretations offer encapsulation and may reduce the computation cost for some modules significantly. The modular approach can play an important role in query optimization, efficient termination, programming design, and software engineering. We classify query types and answer types then discuss query optimization for some query types. Many efficient query processing strategies are applicable to restricted subclasses of programs. The program partition method allows us to select the most efficient strategy for each module. For example, if a module is a uniformly bounded recursion, then the module can be terminated efficiently. If a module defines the transitive closure, then efficient program transformations may be applied to this module