基于GMDH的LSSVM预测模型及其实证

自从Suykens提出新型统计理论学习方法-最小二乘支持向量机(LSSVM)以来,这种方法引起了广泛的关注,它在预测方面的良好性能得到了广泛应用.应用自组织数据挖掘(GMDH)方法改进LSSVM,提升了预测精度.首先利用GMDH方法选择有效的输入变量,再将这些变量作为LSSVM模型的输入,进行时间序列的预测,从而建立LSSVM和GMDH组合的混合模型GLSSVM.并通过汇率时间序列对本文模型进行了实证.结果表明,混合模型预测精度得到了明显的提高.     

正权重组合预测模型及其在经济中的应用

介绍了组合预测的方法,并利用最优组合和递归方差倒数方法对组合预测方法进行改进;提出通过GMDH方法首先对影响经济预测模型的各变量进行筛选然后再建立回归模型、神经网络模型等单项预测模型的思想;最后结合GMDH方法建立的时间序列模型,建立正权重组合预测模型．     

基于GMDH-SVR致密气开发过程参数优化

本文基于非常规油气的致密气开发流程图,利用K-means将4291口气井根据单位压降产气量聚为三类。通过改进GMDH算法对训练集和测试集样本规模比例的完备性和样本次序的随机性进行研究以增强算法鲁棒性。以单位压降产气量作为评价参数对三类气井开发效能主要相关的11个参数进行特征提取并统计各参数作为主成分的概率,选择累计概率80%及以上最少参数作为SVR模型的输入变量,单位压降产气量作为输出变量,选择RBF 作为核函数。分别使三类井各一个样本井的压裂参数上下浮动20%,应用SVR模型统计相应单位压降产气量变化范围是[-18.08%,13.42%]、[-2.34%,5.39%]、[-16.10%,15.21%],分析不同压裂参数组合对单位压降产气量的影响趋势,确定工程实践最优压裂参数,提高气井效能。     

求解全局优化问题的改进蝙蝠算法[英文]

蝙蝠算法(BA)是一类基于试探技巧的群智能优化算法,该算法已被广泛用于诸多领域问题的求解.本文提出一个改进的蝙蝠算法NIBA.在算法中,为了加强蝙蝠算法的局部和全局搜索能力,提出了三个改进策略.首先,为了改进蝙蝠的局部搜索能力,在当前最优解处给出了一个新的搜索方程.其次,为了改进算法的全局搜索能力,平衡算法的开发能力和探索能力,算法吸收并改进了和声搜索机制.最后,为了进一步提高NIBA算法的搜索能力,在当前最优解处,算法采用了混沌搜索机制.为了验证算法的性能,针对18个标准测试函数进行了数值实验.与其它算法的比较结果显示,NIBA算法具有更好的稳定性,且效率更高.     

利用模拟退火遗传算法实现图像阈值分割

本文提出了一种利用模拟退火算法和遗传算法相结合的图像阈值分割算法,试验结果表明该算法增强了算法的全局收敛性,加快了算法的收敛速度,提高了图像阈值分割的效率.     

矩阵填充的硬阈值修正算法

提出了使用硬阈值进行矩阵填充的修正算法.算法通过对迭代矩阵进行对角修正来完成矩阵填充,并给出了算法的收敛性分析.最后通过数值实验比较了修正算法与硬阈值算法填充的数值结果,显示出了新算法的优越性.     

基于乌鸦喝水的元启发式优化算法

启发式优化算法已成为求解复杂优化问题的一种有效方法,可用于解决传统的优化方法难以求解的问题.受乌鸦喝水寓言故事启发,提出一种新型元启发式优化算法—乌鸦喝水算法,首先建立了乌鸦喝水算法数学模型;其次,给出实现该算法的详细步骤;最后,将该算法用于基准函数优化,并将该算法与乌鸦搜索算法、粒子群优化算法、多元宇宙优化算法、花授粉算法、布谷鸟算法等群智能算法进行了比较.仿真实验结果表明,乌鸦喝水算法优于其他算法.     

学习算法的稳定性与泛化(Ⅱ):有界差分稳定框架

根据有界差分条件,提出了学习算法的有界差分稳定框架.依据新框架,研究了机器学习阈值选择算法,再生核Hilbert空间中的正则化学习算法,Ranking学习算法和Bagging算法,证明了对应学习算法的有界差分稳定性.所获结果断言了这些算法均具有有界差分稳定性,从而为这些算法的应用奠定了理论基础.     

一类凸优化的加速混合下降算法

凸优化问题的混合下降算法利用近似条件的已知信息和随机数扩张预测校正步得到了一组下降方向.而前向加速收缩算法利用高斯赛德尔迭代算法的技术,结合邻近点算法和近似邻近点算法的思想,构造了富有扩张性的下降方向.本文借鉴混合下降算法和前向加速收缩算法的思想,利用已有近似规则信息改善了混合下降算法的下降方向,得到了一类凸优化问题的加速混合下降算法.随后利用Markov不等式、凸函数性质和投影的基本性质等,实现了算法的依概率收敛证明.一系列数值试验表明了加速混合下降算法的有效性和效率性.     

基于遗传模拟退火算法的矢量量化码书设计

首先介绍了遗传算法和模拟退火算法等全局优化算法,并针对遗传算法的早熟现象和容易陷入局部最优的缺点,将模拟退火算法引入到遗传算法中,提出了遗传模拟退火矢量量化码书设计(GSAKVQ)算法.此外,针对基于划分的染色体编码方式的特点,算法提出了新的有效的交叉算子和变异算子.同时,将算法从输入空间映射到特征空间,提出了相应的遗传模拟退火核矢量量化算法,改善了算法在某些数据集上的不足.最后,通过实验表明,GSAKVQ算法,在大部分的数据集上都能取得较好的结果,从而验证了算法在数据聚类问题上的有效性.     

A GMDH-based fuzzy modeling approach for constructing TS model

In this paper, a new learning algorithm based on group method of data handling (GMDH) is proposed for the identification of Takagi-Sugeno fuzzy model. Different from existing methods, the new approach, called TS-GMDH, starts from simple elementary TS fuzzy models, and then uses the mechanism of GMDH to produce candidate fuzzy models of growing complexity until the TS model of optimal complexity has been created. The main characteristic of the new approach is its ability to identify the structure of TS model automatically. Experiments on Box-Jenkins gas furnace data and UCI datasets have shown that the proposed method can achieve satisfactory results and is more robust to noise in comparison with other TS modeling techniques such as ANFIS.     

Mathematical model of liquid–liquid equilibrium for a ternary system using the GMDH-type neural network and genetic algorithm

A GMDH type-neural network was used to predict liquid phase equilibrium data for the (water + ethanol + trans-decalin) ternary system in the temperature range of 300.2–315.2 K. In order to accomplish modeling, the experimental data were divided into train and test sections. The data set was divided into two parts: 70% were used as data for “training” and 30% were used as a test set. The predicted values were compared with those of experimental values in order to evaluate the performance of the GMDH neural network method. The results obtained by using GMDH type neural network are in excellent agreement with the experimental results.     

自组织理论和BP人工神经网络在税收预测中的应用研究

针对BP人工神经网络的结构特性,提出了将自组织理论与BP人工神经网络相结合的思想,不仅解决了输入待定的神经网络输入维数难确定的问题,而且加快了神经网络的收敛速度,增强了神经网络的适应能力.并将新建立的模型应用到税收预测中,得出了比常规经济学模型更优的效果.     

GMDH algorithms for complex systems modelling

At present, GMDH algorithms give us a way to identify and forecast economic processes in the case of noised and short input sampling. In contrast to neural networks, the results are explicit mathematical models, obtained in a relatively short time. For ill-defined objects with very big noises, better results are obtained by analog complexing methods. Nets with active neurons should be applied to increase accuracy. Active neurons are able, during the self-organizing process, to estimate which inputs are necessary to minimize the given objective function of the neuron. In the neuronet with such neurons, we have a twofold multilayered structure: neurons themselves are multilayered, and they will be united into a multilayered network.

SelfOrganize! is an easy-to-use modelling tool which realizes twice-multilayered neu-ronets and enables the creation of time series, single input/single output, multi-input/single output and multi-input/multi-output systems (system of equations). Successful applications are shown in the field of analysis and prediction of characteristics of stock markets in financial risk control modelling.     

Cash demand forecasting in ATMs by clustering and neural networks

To improve ATMs’ cash demand forecasts, this paper advocates the prediction of cash demand for groups of ATMs with similar day-of-the week cash demand patterns. We first clustered ATM centers into ATM clusters having similar day-of-the week withdrawal patterns. To retrieve “day-of-the-week” withdrawal seasonality parameters (effect of a Monday, etc.) we built a time series model for each ATMs. For clustering, the succession of seven continuous daily withdrawal seasonality parameters of ATMs is discretized. Next, the similarity between the different ATMs’ discretized daily withdrawal seasonality sequence is measured by the Sequence Alignment Method (SAM). For each cluster of ATMs, four neural networks viz., general regression neural network (GRNN), multi layer feed forward neural network (MLFF), group method of data handling (GMDH) and wavelet neural network (WNN) are built to predict an ATM center’s cash demand. The proposed methodology is applied on the NN5 competition dataset. We observed that GRNN yielded the best result of 18.44% symmetric mean absolute percentage error (SMAPE), which is better than the result of Andrawis, Atiya, and El-Shishiny (2011). This is due to clustering followed by a forecasting phase. Further, the proposed approach yielded much smaller SMAPE values than the approach of direct prediction on the entire sample without clustering. From a managerial perspective, the clusterwise cash demand forecast helps the bank’s top management to design similar cash replenishment plans for all the ATMs in the same cluster. This cluster-level replenishment plans could result in saving huge operational costs for ATMs operating in a similar geographical region.     

Prediction of the paper path in a laser printer by finite element analysis

A conventional problem of paper handling in a page-printing printer is to design a reliable paper path. Sometimes, this also affects the printing quality. The nonlinear theory of elastica has often been used to model paper deflection shape and to predict the paper path between some guiding rollers. The present paper proposes another method that can solve this problem more efficiently. Considering a piece of paper as a nonlinear beam, finite element analysis with geometric nonlinearity and manipulating the contact gap element between paper and guiding surfaces can be used to simulate the large-deflection behavior of paper. In this paper, the paper path in the vicinity of an OPC (organic photoconductive) drum is investigated with a general purpose finite element package. The effect of the electrostatic force between the OPC drum and the paper is considered, and the deformed shape of the paper for different forces lengths is obtained. According to those results an appropriate paper guide is added for an exact paper path.     

A note on “Existence and uniqueness of coexistence states for an elliptic system coupled in the linear part”, by Hei Li-jun,Nonlinear Anal. Real World Appl. 5, 2004

In this short paper I report on a paper published in Nonlinear Analysis: Real World Applications in 2004. There is a major mistake early in that paper which makes most of its claims false. The class of reaction–diffusion systems considered in the paper has been the object of a renewed investigation in the past few years, by myself and others, and recent discoveries provide explicit counter-examples.