GPU implementation of neural networks

Graphics processing unit (GPU) is used for a faster artificial neural network. It is used to implement the matrix multiplication of a neural network to enhance the time performance of a text detection system. Preliminary results produced a 20-fold performance enhancement using an ATI RADEON 9700 PRO board. The parallelism of a GPU is fully utilized by accumulating a lot of input feature vectors and weight vectors, then converting the many inner-product operations into one matrix operation. Further research areas include benchmarking the performance with various hardware and GPU-aware learning algorithms.     

On modelling of curl in multi-ply paperboard

This paper describes a grey-box model for the curl and twist of the carton board produced at AssiDomän Frövi, Sweden. The main equations are based on classical lamination theory of composite materials, and each constituent ply is considered as a macroscopic homogeneous, elastic medium. The model used data from June to September 2004, and shows a general agreement between predicted and measured curvatures. The data were cleaned from outliers by means of the Hampel filter, a nonlinear moving window filter, and with a model based method. Regularization and backwards elimination were used to cope with the low identifiability of the problem. The model was then complemented with a sub-model of immeasurable/unmodelled disturbances estimated with an extended Kalman filter.     

Neural network modelling of word production in Finnish: coding semantic and non-semantic features

The objective of our research is to computationally model word production and its disorders by means of artificial neural networks. In the current study we develop and analyze an algorithm that generates a distributed semantic coding from a given semantic tree-structure classification of words. With the algorithm it is possible to generate semantic representations that are compact and easy to modify. This renders the coding method suitable for our multilayer perceptron-based neural network model of word production. The model is shown to be able to account for a variety of performance patterns observed in four Finnish aphasia patients suffering from word-finding difficulties.     

Evaluating the sustainability of electronic media: Strategies for life cycle inventory data collection and their implications for LCA results

This paper compares two Life Cycle Assessment (LCA) studies independently carried out to assess the environmental impacts of electronic versus print media. Although the two studies lead to the same overall conclusion for the case of a news magazine – namely that the tablet version of the magazine has environmental advantages over the print version – there are significant differences in the details of the LCA results. We show how these differences can be explained by differences in the methodological approaches used for life cycle inventory (LCI) modelling, in particular the use of rough average data versus the attempt to use the most specific and detailed data as possible. We conclude that there are several issues in LCA practice (at least when applied in the domain of media) that can significantly influence the results already at the LCI level: The data collection strategy used (e.g. relying on desk-based research or dismantling a given device) and the decisions made at inventory level with regard to parameters with significant geographic variability, such as the electricity mix or recycling quotas.     

Large-scale simulations of self-organizing neural networks on parallel computers: application to biological modelling

In this contribution we report about a study of a very versatile neural network algorithm known as “Self-organizing Feature Maps” and based on earlier work of Kohonen [1,2]. In its original version, the algorithm addresses a fundamental issue of brain organization, namely how topographically ordered maps of sensory information can be formed by learning.

This algorithm is investigated for a large number of neurons (up to 16 K) and for an input space of dimension d900. To meet the computational demands this algorithm was implemented on two parallel machines, on a self-built Transputer systolic ring and on a Connection Machine CM-2.

We will present below

1. (i) a simulation based on the feature map algorithm modelling part of the synaptic organization in the “hand-region” of the somatosensory cortex,

2. (ii) a study of the influence of the dimension of the input-space on the learning process,

3. (iii) a simulation of the extended algorithm, which explicitly includes lateral interactions, and

4. (iv) a comparison of the transputer-based “coarse-grained” implementation of the model, and the “fine-grained” implementation of the same system on the Connection Machine.

     

Neural networks applications in concrete structures

This paper presents and discusses the applications of neural networks in concrete structures. It aims at introducing neural networks applications in structural design. The paper covers two applications of neural networks in concrete structures. Backpropagation networks are chosen for the proposed network, which is written using the programming package MAT-LAB. The overall results are compared and observed for the performance of the networks. Based on the applications it was found that neural networks are comparatively effective for a number of reasons, which include the amount of CPU memory consumed by neural networks is less than that consumed by conventional methods and their ease of use and implementation, neural networks provide both the users and the developers more flexibility to cope with different kinds of problems.     

Artificial neural network control of a heat exchanger in a closed flow air circuit

This paper experimentally investigates the control of a heat exchanger in a closed flow air circuit. The temperature inside the test section of the test facility has been maintained at a set value by variation of air flow rate over the heat exchanger tube surface and the water flow inside the heat exchanger tubes. The neural network based control has been implemented in a Labview platform and compared with the PID control. The performance of the controller has been investigated for multiple changes in set points and under externally imposed disturbance. The neural network based control has higher speed of response and the steady-state error for the neural network control has a smaller average value than that of the PID control. The control action based on the neural network technique shows less oscillation in comparison to that of the PID based control. Dual actuations, i.e. both air flow and water flow control, have better performance than that with single actuation, i.e. either air flow or water flow control. Both the ANN and PID based control are equally robust in the presence of externally imposed disturbance.     

Neural networks for detecting cyclic behavior in autocorrelated process

In quality control area, cyclic behavior is one of the signals indicating an out-of-control situation in a manufacturing process. Neural network (NN)-based approaches have been proposed to detect the cyclical pattern in the data set collected from process. However, virtually all such proposed methods assume that the process data is independent and identically distributed when the process is under control. In other words, data from a manufacturing process is assumed uncorrelated. In this paper, a NN-based model for detecting the cyclical pattern in an autocorrelated process is proposed. After collecting the process data, this data set is preprocessed using the same information needed to calculate the fractal dimension of the data. It is then fed to a trained feedforward NN with a scaled conjugate gradient backpropagation training algorithm. An output of the model is the state of the process, i.e., whether the process is in-control or out-of-control with a particular cycle period. Such information can assist users of a manufacturing process to identify and remove the underlying causes of the out-of-control state. Our approach is thus suitable for automated manufacturing environment as a supplementary tool to traditional control charts. A Monte Carlo simulation was carried out to study performance of our proposed model. The results showed that the neural-based approach can quickly detect the cyclical pattern with better than 90% accuracy when the signal-to-noise ratio is greater than or equal 2.00. It performs well not only on autocorrelated data for a wide range of autoregressive coefficients, but also on uncorrelated data.     

Bayesian selective combination of multiple neural networks for improving long-range predictions in nonlinear process modelling

A Bayesian selective combination method is proposed for combining multiple neural networks in nonlinear dynamic process modelling. Instead of using fixed combination weights, the probability of a particular network being the true model is used as the combination weight for combining that network. The prior probability is calculated using the sum of squared errors of individual networks on a sliding window covering the most recent sampling times. A nearest neighbour method is used for estimating the network error for a given input data point, which is then used in calculating the combination weights for individual networks. Forward selection and backward elimination are used to select the individual networks to be combined. In forward selection, individual networks are gradually added into the aggregated network until the aggregated network error on the original training and testing data sets cannot be further reduced. In backward elimination, all the individual networks are initially aggregated and some of the individual networks are then gradually eliminated until the aggregated network error on the original training and testing data sets cannot be further reduced. Application results demonstrate that the proposed techniques can significantly improve model generalisation and perform better than aggregating all the individual networks.     

基于神经网络的油田注水泵站优化控制

针对油田注水系统能耗大的问题,提出了一种以系统效率模型为依据的注水泵站优化控制方法.本方法可以改善能耗模型中系统总是运行在控制扬程高限的问题,进一步提高系统效率,降低能耗.泵站同时使用离心泵和往复泵,运用神经网络技术,建立了多台水泵并联运行的优化调度数学模型,结合完全枚举法和复合形法给出了其求解算法.仿真实验表明该方法是有效的.     

Intelligent non-linear modelling of an industrial winding process using recurrent local linear neuro-fuzzy networks

This study deals with the neuro-fuzzy (NF) modelling of a real industrial winding process in which the acquired NF model can be exploited to improve control performance and achieve a robust fault-tolerant system. A new simulator model is proposed for a winding process using non-linear identification based on a recurrent local linear neuro-fuzzy (RLLNF) network trained by local linear model tree (LOLIMOT), which is an incremental tree-based learning algorithm. The proposed NF models are compared with other known intelligent identifiers, namely multilayer perceptron (MLP) and radial basis function (RBF). Comparison of our proposed non-linear models and associated models obtained through the least square error (LSE) technique (the optimal modelling method for linear systems) confirms that the winding process is a non-linear system. Experimental results show the effectiveness of our proposed NF modelling approach.     

延迟离散神经网络动态特征的矩阵判据

该文利用延迟离散网络的状态转移方程与矩阵不等式的等价性研究网络的动力学性质.对于具有任意连接权阵的网络,文章给出了周期为1和2的极限环存在的一些条件.同时,对于周期为1,2和4的一些特殊极限环给出了存在的条件,还得到了网络不存在任何不动点的充分条件,即只有极限环的充分条件.计算机模拟实验表明其结果是正确的.     

Product-service system engineering characteristics design for life cycle cost based on constraint satisfaction problem and Bayesian network

A product-service system (PSS) has many engineering characteristics (ECs), their design is a critical work in PSS planning, which has an important influence on the cost and quality of PSS. How to design a reasonable PSS-ECs scheme, and evaluate its life cycle cost (LCC) is a challenging task. Aiming at the PSS-ECs design for LCC, this paper proposes a new PSS design method, it first treats and models the design of PSS-ECs as a customer requirements-based constraint satisfaction problem (CSP) for finding an initial set of satisfied PSS-ECs schemes, and then it evaluates these schemes based on Bayesian network (BN)-based LCC estimation model for finding an optimal scheme as a solution. Constructing a BN describing the uncertain relationships between PSS-ECs and LCC is the core of this research. By combining existing R&D data and expert experience, Bayesian estimation and arithmetic averaging are used to estimate the conditional probability in BN. Take a subway bogie and its maintenance service in a Chinese company as an example to verify the proposed method. The results show that the proposed method can effectively solve the problem of PSS-ECs design for LCC, it also shows that this method has positive significance in realizing engineering knowledge consolidation, assisting designers in exploring design space, and improving the rationality of design decisions.     

Neural network prediction of relapse in breast cancer patients

When a woman diagnosed as having breast cancer has a tumour removed, it is important to try and predict whether she is likely to relapse within, say, the next three years. In this paper, the performance of a neural network classifier trained on a number of prognostic indicators is shown to be better than that of the clinical experts working with the same information. To obtain meaningful statistics with the relatively small dataset available, the network is trained using a modified form of the leave-one-out method. A procedure is also introduced for investigating how much independentinformation each input parameter contributes. This shows that, in this type of retrospective study, the type of therapy given to the woman does not significantly affect the network's prediction of whether or not she will relapse within three years. Finally, since this problem, in common with many other medical problems, is plagued by a shortage of data, the final section of the paper reports on an investigation of whether or not multi-centre databases might be feasible.     

Artificial neural networks for modeling in reaction process systems

This work is focused on developing a more efficient computational scheme for estimation of process reaction rates based on NN models. Two scenarios are considered: (1) the kinetics coefficients of the process are completely known and the process states are partly known (measured); (2) the kinetics coefficients and the states of the process are partly known. The contribution of the paper is twofold. From one side we formulate a hybrid (ANN and mechanistic) model that outperforms the traditional reaction rate estimation approaches. From other side, a new procedure for NN supervised training is proposed when target outputs are not available. The two scenarios are successfully tested for two benchmark problems, estimation of the precipitation rate of calcium phosphate and estimation of sugar crystallization growth rate.     

神经网络在注塑制品材料选择中的应用

该文分析了注塑成型制品的功能要求与树脂材料性质之间的作用关系，采用模糊手段表示了对材料的选定度，提出了一种基于神经网络的注塑制品的材料选择方法。     

Neural networks for bandwidth selection in local linear regression of time series

The problem of automatic bandwidth selection in nonparametric regression is considered when a local linear estimator is used to derive nonparametrically the unknown regression function. A plug-in method for choosing the smoothing parameter based on the use of the neural networks is presented. The method applies to dependent data generating processes with nonlinear autoregressive time series representation. The consistency of the method is shown in the paper, and a simulation study is carried out to assess the empirical performance of the procedure.     

遗传算法和神经网络在白细胞自动识别中的应用

该文讨论并建立了一真彩色白细胞显微图像微机自动识别系统。针对传统组合优化方法用于特征选择的种种缺陷,提出了基于遗传算法的特征选择策略。根据细胞特征,笔者对传统的遗传算法做了相应的改进。最后运用选择特征子集来训练BP神经网络作为分类器。结果证明该方法对于分类系统的性能优化有明显效果。     

VAV系统送风管道静压神经元网络控制的仿真研究

变风量(variable air volume,VAV)空调系统中送风管道静压的控制效果直接决定了VAV系统的稳定性和节能能力。但送风管道静压的强耦合特性往往使得其控制效果很不理想。针对这种情况,该文利用了神经元网络具有自学习以及超强非线性逼近的能力,提出了基于线性神经元网络的补偿控制方法。这种控制方法能够在线学习,根据送风管道静压耦合因素的变化自适应的调节控制量,实现对管道静压的补偿控制。文中给出了神经元权系数的在线学习公式,建立了被控对象数学模型,并通过Matlab仿真验证了控制算法的有效性。     

基于模糊聚类的神经网络求解围棋死活博弈

围棋死活求解非常消耗计算资源,博弈树的节点数随分支因子和深度的增加而呈指数级增长,使得传统的完全遍历博弈树的搜索不足以胜任.文中提出了一种基于模糊聚类的神经网络方法,利用模式识别和模糊属性检测,为涉及到外部劫争、循环规避、哈希置换、证明树等多方面问题的围棋博弈死活求解,构建了基于神经网络的棋型聚类分析器,快速又有效地极大减小了博弈树的分支因子,节约了死活求解所耗费的计算时间和内存空间.实验证明取得了比较理想的结果,研究表明把自学习能力赋予程序是提高计算机围棋博弈性能的有效途径.