汇率货币模型的非线性协整关系检验——基于深度GRU神经网络

本文采用深度门控循环单元（GRU）神经网络探讨三种汇率货币模型（弹性价格、前瞻性和实际利率差模型）的非线性协整关系。GRU技术在深度学习中具有智能记忆、自主学习和强逼近能力等优点。为此，本文运用该技术对6组典型浮动汇率制国别数据进行了非线性Johansen协整检验。结果表明，汇率与宏观经济基本面之间存在非线性协整关系，从而说明了货币模型在非线性条件下的有效性，以及先进的深度学习工具在检验经济理论中的优势。     

A noise-based stabilizer for convolutional neural networks

Overfitting occurs when one tries to train a large model on small amount of data. Regularizing a neural network using prior knowledge remains a topic of research as it is not concluded how much prior information can be given to the neural network. In this paper, a novel algorithm is introduced which uses regularization to train a neural network without increasing the dataset. A trivial prior information of a class label is supplied to the model while training. Laplace noise is introduced to the intermediate layer for more generalization. The results show significant improvement in accuracy on the standard datasets for a simple Convolutional Neural Network (CNN). While the proposed method outperforms previous regularization techniques like dropout and batch normalization, it can also be applied with them for further improvement in the performance. On the variants of MNIST, proposed algorithm achieved an average 48% increment in the test accuracy.     

交流电机变绕组匝数拓宽调速范围的新方法

在典型交流电机调速应用领域需要拓宽电机的调速范围。传统的拓宽电机调速范围的方法主要包括弱磁调速、 变极调速以及改变绕组结构等。针对电动汽车驱动电机应用背景,提出了一种拓宽交流电机调速范围的新方法,即在维 持电机本体结构与变频器功率等级基本不变的情况下,利用由少量全控和不控电力电子器件构成的绕组调节功率变换 电路,实现电机定子绕组等效匝数的连续可调。仿真结果表明,定子绕组等效匝数是可以通过脉冲控制方法实现连续可 调的,应用该方法,可以拓宽电机的调速范围,使其适应不同应用场合对电机转矩曲线的要求。     

Completing the Circuit: Routine,Reflection, and Ethical Consumption

Efforts to explain why some people incorporate ethical concerns into everyday shopping for food and household goods, while many do not, have so far left significant variation in “ethical consumption” unexplained. Seeking to move beyond explanations that rely mainly on differences in consumers' social class, gender, and political engagement, I draw on concepts associated with “practice theory” to argue that ethical consumption is closely tied to people's willingness and ability to spend time, while shopping, on distinct activities associated with breaking old routines and establishing new ones. The central insight of practice theory is that most consumption is the product of unconscious routine. And it is precisely because consciously departing from routine is, according to my study, a fundamentally time‐consuming process, that lack of time emerges as a crucial obstacle to translating abstract ethical concerns into concrete action as a consumer.     

党政机构微博在突发事件中传播效果的影响因素及策略

在分析铁路货运量预测方法的基础上，针对标准BP神经网络的不足，提出改进的BP神经网络预测模型。首先,利用动态陡度因子来改变激励函数的陡峭程度，以此来得到更好的激励函数响应特征以及更好的非线性表达能力；其次，利用附加动量因子，通过对以前经验的积累，既降低了神经网络对误差曲面的局部细节敏感特性，又较好的遏制了神经网络易于限于局部最小的缺陷；最后，采取改变学习率的方法，给定一个较大的学习率初始值，在学习的过程中学习率不断减小，网络最终趋于稳定。改进BP算法既可以得到更优的解，还能够缩短训练时间。利用全国铁路货运     

Bootstrap order selection for SETAR models

In this paper, we investigate the selecting performances of a bootstrapped version of the Akaike information criterion for nonlinear self-exciting threshold autoregressive-type data generating processes. Empirical results will be obtained via Monte Carlo simulations. The quality of our method is assessed by comparison with its non-bootstrap counterpart and through a novel procedure based on artificial neural networks.     

Back propagation neural networks and multiple regressions in the case of heteroskedasticity

This paper compares the performance between regression analysis and a clustering based neural network approach when the data deviates from the homoscedasticity assumption of regression. Heteroskedasticity is a problem that arises in linear regression due to the unequal error variances. One of the methods to deal heteroskedasticity in classical regression theory is weighted least-square regression (WLS). In order to deal the problem of heteroskedasticity, backpropagation neural network is applied. In this context, an algorithm is proposed which is based on robust estimates of location and dispersion matrix that helps in preserving the error assumption of the linear regression. Analysis is carried out with appropriate designs using simulated data and the results are presented.     

基于盾构掘进参数的LVQ神经网络地层识别

以苏州4号线2标及2号线东延伸线5标地铁工程为背景，分析了盾构机的掘进参数：千斤顶推力、推进速度、刀盘扭矩、螺旋机转速和同步注浆量在不同地层条件下的变化规律。提出了基于盾构机掘进参数的学习向量量化（Learning Vector Quantization，LVQ）神经网络地层识别方法。建立了以盾构机五个掘进参数作为输入，地层特性编码为输出的数学模型，通过每种地层100组训练样本对模型进行训练，通过57步训练，训练样本误差控制在0.1以内，并用每种地层50组检验样本进行检验，地层总体识别率达到82.7%。     

基于概率神经网络和K-means算法的纳税评估

纳税评估工作是一项难于建立准确数学模型的复杂系统,同时又是一个典型的模式识别问题。用神经网络方法进行纳税评估有其独特的优越性。运用PNN算法很大程度地依赖于训练样本对象的选取。选取的这些样本能否反映总体的信息特征决定了分类器的识别效果。文章运用K-means算法对纳税人信息样本进行聚类,找出聚类中心点,以此为基础来选择样本作为PNN的训练样本,从而达到对PNN算法的优化。研究结果表明这种改进后的PNN算法分类效果好,对于纳税评估有其应用价值。     

Nonlinear models of disability and age applied to census data

It is usually considered that the proportion of handicapped people grows with age. Namely, the older the man/woman, the more the level of disability he/she suffers. However, empirical evidence shows that this assessment is not always true, or at least, it is not true in the Spanish population. The study tries to assess the impact of age on disability in Spain. Each gender has been treated separately because it can be shown that men and women have their own pattern of behaviour. Three different methods of estimation have been used to check the link between those variables. The results seem to support the idea that the relationship among age and the intensity of disability is not always direct. One of the concluding remarks in this analysis is that the method of estimation has a great incidence in the final results, especially in central ages between 20 and 80 years old.