Can purchasing power parity help forecast the dollar?

This paper examines the information on future exchange rate movements provided by the doctrine of purchasing power parity (PPP). Previous research has studied this issue by analyzing the time-series properties of period-by-period levels of, or changes in, exchange rates. In contrast, the present study focuses on the durations of periods in which exchange rates deviate from their PPP levels. If PPP provides information about future exchange rate movements, these durations should exhibit positive duration dependence. That is, the probability of returning to PPP levels should increase as the period of deviation increases. Parametric hazard functions estimated using data from eighteen countries provide no evidence of positive duration dependence. These results are robust to alternative definitions of PPP and to alternative functional specifications. While exchange rates take prolonged swings away from their PPP levels and then eventually return, these movements apparently constitute Monte Carlo cycles in which, at any point in time, the probability of moving back toward PPP is the same as the probability of moving farther away. Thus, PPP provides no useful information on future exchange rate changes, a result consistent with market efficiency.     

一种结构自适应的聚类神经网络

为了解决聚类分析中聚类数的确定问题,在SOFM神经网络的基础上,从聚类准则出发,通过试验对聚类准则的曲线特征进行了详细的分析和论证,设计出一种结构自适应的聚类神经网络,该网络能自动确定最佳的聚类数,并提出了一种减少计算量的改进算法。     

日径流预报贝叶斯回声状态网络方法

回声状态网络（ESN）相比传统递归神经网络,具有模型简单、参数训练速度快的特点．针对标准ESN因常采用线性回归率定模型参数容易出现过拟合问题,提出了基于贝叶斯回声状态网络（BESN）的日径流预报模型．该模型将贝叶斯理论与ESN模型相结合,通过权重后验概率密度最大化而获得最优输出权重,提高了模型的泛化能力．通过安砂和新丰江两座水库日径流预测实例表明,BESN模型是一种有效、可行的预测方法,与传统BP神经网络和ESN模型对比,进一步表明BESN模型具有更好的预测精度．     

互联网时代语音识别基本问题

语音识别技术经过半个世纪的积累,于近年来达到大规模商用水平．本文概括了统计语音识别理论的发展状况,并单独介绍了深度神经网络在声学建模、语言建模、多语言共享、语义识别等方面的卓越性能．深度神经网络的性能优势引起了我们强烈的兴趣．通过回顾类人听觉信息处理对深度神经网络的改进作用,我们意识到,深度神经网络与类人听觉信息处理相结合,必将推进语音识别技术的进一步发展．反过来,深度神经网络技术在语音识别中的进步,也必将推动类人听觉信总、处理技术的进步．语音识别技术后续发展的重点是对深度神经网络的结构和训练算法的改进使之更好地实现类人听觉．最后,我们分析了采用深度神经网络模拟人类听觉的抗噪修复机理和听觉关注机理的可能性．     

Performance evaluation of neural network architectures: the case of predicting foreign exchange correlations

In the last decade, neural networks have emerged from an esoteric instrument in academic research to a rather common tool assisting auditors, investors, portfolio managers and investment advisors in making critical financial decisions. It is apparent that a better understanding of the network's performance and limitations would help both researchers and practitioners in analysing real‐world problems. Unlike many existing studies which focus on a single type of network architecture, this study evaluates and compares the performance of models based on two competing neural network architectures, the multi‐layered feedforward neural network (MLFN) and general regression neural network (GRNN). Our empirical evaluation measures the network models' strength on the prediction of currency exchange correlation with respect to a variety of statistical tests including RMSE, MAE, U statistic, Theil's decomposition test, Henriksson–Merton market timing test and Fair–Shiller informational content test. Results of experiments suggest that the selection of proper architectural design may contribute directly to the success in neural network forecasting. In addition, market timing tests indicate that both MLFN and GRNN models have economically significant values in predicting the exchange rate correlation. On the other hand, informational content tests discover that the neural network models based on different architectures capture useful information not found in each other and the information sets captured by the two network designs are independent of one another. An auxiliary experiment is developed and confirms the possible synergetic effect from combining forecasts made by the two different network architectures and from incorporating information from an implied correlation model into the neural network forecasts. Implied correlation and random walk models are also included in our empirical experiment for benchmark comparison. Copyright © 2005 John Wiley & Sons, Ltd.     

Forecast combining with neural networks

This paper investigates the use of Artificial Neural Networks (ANNs) to combine time series forecasts of stock market volatility from the USA, Canada, Japan and the UK. We demonstrate that combining with nonlinear ANNs generally produces forecasts which, on the basis of out-of-sample forecast encompassing tests and mean squared error comparisons, routinely dominate forecasts from traditional linear combining procedures. Superiority of the ANN arises because of its flexibility to account for potentially complex nonlinear relationships not easily captured by traditional linear models.     

基于神经网络的Q学习在Khepera Ⅱ机器人避障中的应用

为了提高移动机器人的自主学习能力,在传统的机器人行为控制结构基础上设计了智能控制结构,同时引入了基于神经网络的Q学习模块算法,克服了传统算法只能应用到离散状态中的不足.移动机器人的避障实验结果表明,该方法能够使移动机器人通过自学习实现自主避障.     

BP神经网络样本数据预处理应用研究

提出一种新的线性预处理输入数据的方法,即通过线性运算将样本数据的各个字段值统一到同一个数量级,然后结合数值归一化的方法将数据运用到神经网络。在基于信令的漫游用户实时信用度测评及欠费风险超前控制系统中,使用统一字段值的数量级的方法预处理样本数据取得了很好的预测效果。由此,在模式识别和预测领域,统一样本数据的各个字段值的数量级后再进行网络训练可以取得更好的训练效果。