E-R modeling and visualization of large mutual fund data

Existing visualization techniques used by mutual fund managers focus on the active portfolio management strategies. However, an important aspect of mutual funds is to visualize and understand the fund composition observed at specific points in time. Such a visualization will leverage the fund manager with the power to investigate and to capture forecasting differences and future performance. Our research here is to design a new visualization system that has an Entity-Relationship (E-R) model at the heart of the system and reveals not only the aggregate behavior of the mutual funds, but also the behavior of the mutual fund managers who manage these funds. We find some preliminary evidence that mutual funds in the world market represent a strong geographical pattern, and some potential niche markets are now extracted that stands out in the global scenario such as the returns from funds that belong to a specific country. For the analysis of security distribution, it helps to distinguish the most popular securities from other ordinary ones for an increased realization of profit. Also based on our experience with such huge world fund data to achieve a trade-off between time efficiency and graphics quality, we recommend the approach of splatting visualization, which indicates the influence of a particular security in a fund’s portfolio.     

Modeling electricity spot and futures price dependence: A multifrequency approach

Electricity prices are known to exhibit multifractal properties. We accommodate this finding by investigating multifractal models for electricity prices. In this paper we propose a flexible Copula-MSM (Markov Switching Multifractal) approach for modeling spot and weekly futures price dynamics. By using a conditional copula function, the framework allows us to separately model the dependence structure, while enabling use of multifractal stochastic volatility models to characterize fluctuations in marginal returns. An empirical experiment is carried out using data from Nord Pool. A study of volatility forecasting performance for electricity spot prices reveals that multifractal techniques are a competitive alternative to GARCH models. We also demonstrate how the Copula-MSM model can be employed for finding optimal portfolios, which minimizes the Conditional Value-at-Risk.     

Random matrix theory and fund of funds portfolio optimisation

The proprietary nature of Hedge Fund investing means that it is common practise for managers to release minimal information about their returns. The construction of a fund of hedge funds portfolio requires a correlation matrix which often has to be estimated using a relatively small sample of monthly returns data which induces noise. In this paper, random matrix theory (RMT) is applied to a cross-correlation matrix C, constructed using hedge fund returns data. The analysis reveals a number of eigenvalues that deviate from the spectrum suggested by RMT. The components of the deviating eigenvectors are found to correspond to distinct groups of strategies that are applied by hedge fund managers. The inverse participation ratio is used to quantify the number of components that participate in each eigenvector. Finally, the correlation matrix is cleaned by separating the noisy part from the non-noisy part of C. This technique is found to greatly reduce the difference between the predicted and realised risk of a portfolio, leading to an improved risk profile for a fund of hedge funds.     

Modeling the net flows of U.S. mutual funds with stochastic catastrophe theory

Using the recent work of Hartelman, van der Maas, and Wagenmakers, we demonstrate the use of invariant stochastic catastrophe models in finance for modeling net flows (the difference between purchases and redemptions of fund shares) of U.S. mutual funds. We validate Goetzmann et al. and others' work concerning the importance of sentiment variables on stock fund flows. We also answer some of the questions Goetzmann et al. and Brown et al. pose at the end of their respective papers. We end with possible experiments for experimental economists and sociophysicists.     

A tailored solver for bifurcation analysis of ocean-climate models

In this paper, we present a new linear system solver for use in a fully-implicit ocean model. The new solver allows to perform bifurcation analysis of relatively high-resolution primitive-equation ocean-climate models. It is based on a block-ILU approach and takes special advantage of the mathematical structure of the governing equations. In implicit models Jacobian matrices have to be constructed. Analytical construction is hard for complicated but more realistic representations of mixing. This is overcome by evaluating the Jacobian in part numerically. The performance of the new implicit ocean model is demonstrated using (i) a high-resolution model of the wind-forced double-gyre flow problem in a (relatively small) midlatitude spherical basin, and (ii) a medium-resolution model of thermohaline and wind-driven flows in an Atlantic size single-hemispheric basin.     

A Hyperspectral Image Classification Approach Based on Feature Fusion and Multi-Layered Gradient Boosting Decision Trees

At present, many Deep Neural Network (DNN) methods have been widely used for hyperspectral image classification. Promising classification results have been obtained by utilizing such models. However, due to the complexity and depth of the model, increasing the number of model parameters may lead to an overfitting of the model, especially when training data are insufficient. As the performance of the model mainly depends on sufficient data and a large network with reasonably optimized hyperparameters, using DNNs for classification requires better hardware conditions and sufficient training time. This paper proposes a feature fusion and multi-layered gradient boosting decision tree model (FF-DT) for hyperspectral image classification. First, we fuse extended morphology profiles (EMPs), linear multi-scale spatial characteristics, and nonlinear multi-scale spatial characteristics as final features to extract both special and spectral features. Furthermore, a multi-layered gradient boosting decision tree model is constructed for classification. We conduct experiments based on three datasets, which in this paper are referred to as the Pavia University, Indiana Pines, and Salinas datasets. It is shown that the proposed FF-DT achieves better performance in classification accuracy, training conditions, and time consumption than other current classical hyperspectral image classification methods.     

Rating the raters in a mixed model: An approach to deciphering the rater reliability

Rating the raters has attracted extensive attention in recent years. Ratings are quite complex in that the subjective assessment and a number of criteria are involved in a rating system. Whenever the human judgment is a part of ratings, the inconsistency of ratings is the source of variance in scores, and it is therefore quite natural for people to verify the trustworthiness of ratings. Accordingly, estimation of the rater reliability will be of great interest and an appealing issue. To facilitate the evaluation of the rater reliability in a rating system, we propose a mixed model where the scores of the ratees offered by a rater are described with the fixed effects determined by the ability of the ratees and the random effects produced by the disagreement of the raters. In such a mixed model, for the rater random effects, we derive its posterior distribution for the prediction of random effects. To quantitatively make a decision in revealing the unreliable raters, the predictive influence function (PIF) serves as a criterion which compares the posterior distributions of random effects between the full data and rater-deleted data sets. The benchmark for this criterion is also discussed. This proposed methodology of deciphering the rater reliability is investigated in the multiple simulated and two real data sets.     

Decision Rules Construction: Algorithm Based on EAV Model

In the paper, an approach for decision rules construction is proposed. It is studied from the point of view of the supervised machine learning task, i.e., classification, and from the point of view of knowledge representation. Generated rules provide comparable classification results to the dynamic programming approach for optimization of decision rules relative to length or support. However, the proposed algorithm is based on transformation of decision table into entity–attribute–value (EAV) format. Additionally, standard deviation function for computation of averages’ values of attributes in particular decision classes was introduced. It allows to select from the whole set of attributes only these which provide the highest degree of information about the decision. Construction of decision rules is performed based on idea of partitioning of a decision table into corresponding subtables. In opposite to dynamic programming approach, not all attributes need to be taken into account but only these with the highest values of standard deviation per decision classes. Consequently, the proposed solution is more time efficient because of lower computational complexity. In the framework of experimental results, support and length of decision rules were computed and compared with the values of optimal rules. The classification error for data sets from UCI Machine Learning Repository was also obtained and compared with the ones for dynamic programming approach. Performed experiments show that constructed rules are not far from the optimal ones and classification results are comparable to these obtained in the framework of the dynamic programming extension.     

基于属性约简的应用服务器优化算法改进

优化参数配置是优化应用服务器性能的重要方面;基于传统参数调节的优化策略耗时耗力缺乏系统性和规律性;利用模块化思想针对目标决策函数对应用服务器参数进行分类,可构建条件属性约简模型;基于属性约简的应用服务器优化算法,可去除对于目标决策函数相对不重要的参数,并获得相对重要的参数,从而达到锁定目标重点调节,快速提高系统性能的目的;现有的约简模型优化算法多基于经典粗糙集理论,在等价关系的基础上构造分类,容易造成大量的信息破坏和流失;文章通过拓展等价关系到一般二元关系,利用广义粗糙集理论改良了基于模块化思想和属性约简模型的应用服务器优化算法,通过定义辨识函数对条件属性进行约简,再结合依赖度计算,得到最终目标参数。     

The string prediction models as invariants of time series in the forex market

In this paper we apply a new approach of string theory to the real financial market. The models are constructed with an idea of prediction models based on the string invariants (PMBSI). The performance of PMBSI is compared to support vector machines (SVM) and artificial neural networks (ANN) on an artificial and a financial time series. A brief overview of the results and analysis is given. The first model is based on the correlation function as invariant and the second one is an application based on the deviations from the closed string/pattern form (PMBCS). We found the difference between these two approaches. The first model cannot predict the behavior of the forex market with good efficiency in comparison with the second one which is, in addition, able to make relevant profit per year. The presented string models could be useful for portfolio creation and financial risk management in the banking sector as well as for a nonlinear statistical approach to data optimization.     

Entropy Based Student’s t-Process Dynamical Model

Volatility, which represents the magnitude of fluctuating asset prices or returns, is used in the problems of finance to design optimal asset allocations and to calculate the price of derivatives. Since volatility is unobservable, it is identified and estimated by latent variable models known as volatility fluctuation models. Almost all conventional volatility fluctuation models are linear time-series models and thus are difficult to capture nonlinear and/or non-Gaussian properties of volatility dynamics. In this study, we propose an entropy based Student’s t-process Dynamical model (ETPDM) as a volatility fluctuation model combined with both nonlinear dynamics and non-Gaussian noise. The ETPDM estimates its latent variables and intrinsic parameters by a robust particle filtering based on a generalized H-theorem for a relative entropy. To test the performance of the ETPDM, we implement numerical experiments for financial time-series and confirm the robustness for a small number of particles by comparing with the conventional particle filtering.     

A robust decentralized reputation management system for service selection

Reputation mechanism is a novel approach to automate QoS-aware service selection in service oriented computing. The reputation system collects ratings on QoS that consumers feedback and aggregates them to derive a reputation value, which can in turn assist other consumers in service selection in future. However, current approaches fail to combat the malicious ratings and hence the calculated reputation values can be biased severely or even manipulated. Moreover, the centralized management of rating data restricts its application to large open environment. In this paper, we present a robust decentralized reputation system which can resist various unfair ratings and manipulation behaviours. It can evolve and become more mature against malicious ratings with the system running continuously. At last, we experimentally verify the robustness of the proposed approach through a simulation study.     

复杂装备退化状态早期识别方法研究

针对复杂装备早期退化状态难以识别的问题,提出一种将相关向量机(RVM)和Dezert-Smarandache 理论(DSmT)相结合的多特征融合决策识别方法。该方法首先分别采用时域分析法和时频域小波包变换法对装备的状态特征进行提取;之后将状态特征向量输入RVM模型中完成对状态属性的判定并获得各种状态模式的基本置信度分配;最后依据DSmT的PCR6规则对含有冲突信息的多个识别结果进行决策融合,得到早期退化状态的最终识别结果。在对某航空机电设备的实例应用中表明,该方法可以有效地解决信息高冲突条件下的早期退化状态识别问题,结果可靠准确。     

How to Promote the Performance of Parametric Volatility Forecasts in the Stock Market? A Neural Networks Approach

This study uses the fourteen stock indices as the sample and then utilizes eight parametric volatility forecasting models and eight composed volatility forecasting models to explore whether the neural network approach and the settings of leverage effect and non-normal return distribution can promote the performance of volatility forecasting, and which one of the sixteen models possesses the best volatility forecasting performance. The eight parametric volatility forecasts models are composed of the generalized autoregressive conditional heteroskedasticity (GARCH) or GJR-GARCH volatility specification combining with the normal, Student’s t, skewed Student’s t, and generalized skewed Student’s t distributions. Empirical results show that, the performance for the composed volatility forecasting approach is significantly superior to that for the parametric volatility forecasting approach. Furthermore, the GJR-GARCH volatility specification has better performance than the GARCH one. In addition, the non-normal distribution does not have better forecasting performance than the normal distribution. In addition, the GJR-GARCH model combined with both the normal distribution and a neural network approach has the best performance of volatility forecasting among sixteen models. Thus, a neural network approach significantly promotes the performance of volatility forecasting. On the other hand, the setting of leverage effect can encourage the performance of volatility forecasting whereas the setting of non-normal distribution cannot.     

Simplified hearing protector ratings—an international comparison

A computer was programmed to model the distributions of dB(A) levels reaching the ears of an imaginary workforce wearing hearing protectors selected on the basis of either octave band attenuation values or various simplified ratings in use in Australia, Germany, Poland, Spain or the U.S.A. Both multi-valued and single-valued versions of dB(A) reduction and sound level conversion ratings were considered. Ratings were compared in terms of precision and protection rate and the comparisons were replicated for different samples of noise spectra (N = 400) and hearing protectors (N = 70) to establish the generality of the conclusions. Different countries adopt different approaches to the measurement of octave band attenuation values and the consequences of these differences were investigated. All rating systems have built-in correction factors to account for hearing protector performance variability and the merits of these were determined in the light of their ultimate effects on the distribution of dB(A) levels reaching wearers' ears. It was concluded that the optimum rating is one that enables the dB(A) level reaching wearers to be estimated by subtracting a single rating value from the dB(C) level of the noise environment, the rating value to be determined for a pink noise spectrum from mean minus one standard deviation octave band attenuation values with further protection rate adjustments being achieved by the use of a constant correction factor.     

Reconstructing Nonparametric Productivity Networks

Network models provide a general representation of inter-connected system dynamics. This ability to connect systems has led to a proliferation of network models for economic productivity analysis, primarily estimated non-parametrically using Data Envelopment Analysis (DEA). While network DEA models can be used to measure system performance, they lack a statistical framework for inference, due in part to the complex structure of network processes. We fill this gap by developing a general framework to infer the network structure in a Bayesian sense, in order to better understand the underlying relationships driving system performance. Our approach draws on recent advances in information science, machine learning and statistical inference from the physics of complex systems to estimate unobserved network linkages. To illustrate, we apply our framework to analyze the production of knowledge, via own and cross-disciplinary research, for a world-country panel of bibliometric data. We find significant interactions between related disciplinary research output, both in terms of quantity and quality. In the context of research productivity, our results on cross-disciplinary linkages could be used to better target research funding across disciplines and institutions. More generally, our framework for inferring the underlying network production technology could be applied to both public and private settings which entail spillovers, including intra- and inter-firm managerial decisions and public agency coordination. This framework also provides a systematic approach to model selection when the underlying network structure is unknown.     

A neural networks based model of inverse hysteresis

A novel and simple approach based on transformation using neural networks is proposed in this paper to model the inverse behavior of hysteresis. In this approach, a continuous transformation is used to construct an elementary inverse hysteresis operator (EIHO), which can extract the change tendency of inverse hysteresis. Then based on the EIHO, an expanded input space is constructed to transform the multi-valued mapping of inverse hysteresis into a one-to-one mapping. Based on the constructed expanded input space, a neural network is employed to approximate the inverse hysteresis. Both experiment and simulation are implemented to validate the effectiveness of the proposed approach. These results indicate that the proposed approach has derived satisfactory modeling performance.     

生成式对抗网络的土壤有机质高光谱估测模型

已有的土壤有机质含量估测模型大多以光谱特征波段、线性和非线性模型为基础,较少考虑通过拓展样本数据建模集来提高模型的估测能力。为进一步提高土壤有机质高光谱反演模型估测精度,提出利用生成式对抗网络(GAN)合成伪高光谱数据和有机质含量的动态估测模型。选取湖南省长沙市及周边区域的水稻田为研究对象,采集土样和实测高光谱数据(350~2 500 nm),室内化学测定有机质含量。以高光谱数据和有机质含量为基础,利用生成式对抗网络生成等量新数据, 结合原始数据建模集组成增强建模集。在GAN正式训练中,每轮训练完成后,设置4个观测点(对应增强建模集中含50,100,150和239个生成样本),动态构建交叉验证岭回归(RCV)、偏最小二乘回归(PLSR)和BP神经网络(BPNN)土壤有机质含量估测模型(分别简称GAN-RCV,GAN-PLSR和GAN-BPNN),并在相同测试集上实施模型评估。实验结果表明：(1)原始数据建模集上拟合的估测模型中,交叉验证岭回归表现最佳,决定系数(R2)和均方根误差(RMSE)分别为0.831 1和0.189 6;(2)GAN的150轮正式训练中,增强建模集上动态构建的GAN-RCV,GAN-PLSR和GAN-BPNN模型性能显著提高,具体表现为：GAN-RCV的R2取得最大值0.890 9(RMSE 0.153 7)、最小值0.850 5 (RMSE 0.18)与平均值0.868 7(RMSE 0.168 6),最大R2比建模集上拟合的RCV提高了7.2%(RMSE降低了18.9%),GAN-PLSR获得R2最大值0.855 4(RMSE 0.176 9)、最小值0.727 0 (RMSE 0.243 2)与平均值0.780 1 (RMSE 0.217 7),最大R2比建模集上拟合的PLSR提高了20.6%(RMSE降低了29.5%),GAN-BPNN表现最佳,R2取得最大值0.905 2(RMSE 0.143 3)、最小值0.801 7(RMSE 0.207 3)与平均值0.868 1(RMSE 0.168 6),最大R2比建模集上拟合的BPNN提高了30.8%(RMSE降低了44.5%);(3)随着增强建模集中生成样本数量增加,模型精度提升效果呈先升后降趋势,4个观测点中第3个观测点的模型性能提升最显著。充分的实验表明：基于GAN动态构建的有机质含量估测模型显著改善了模型预测性能。依据测试集上的评估结果,可择优使用最佳模型进行后续土壤有机质含量估测。     

多特征融合的鸟类物种识别方法

针对短时窗平均/长时窗平均算法从次声台站监测数据中提取的信号仍然包含噪声的问题,对支持向量机和人工神经网络的机器学习方法进行了研究。采用小波包分解的方法对信号进行重构,提取出各频带内的重构信号能量特征,对事件信号和噪声进行了识别实验,并分析了提高识别能力的方法,为工程应用提供理论参考。实验结果表明,在训练数据集不大的情况下,通过优化模型结构可以将两种方法的识别能力提高到可以接受的水平。     

On the relative importance of spatial and timbral fidelities in judgments of degraded multichannel audio quality

Mean opinion score ratings of reproduced sound quality typically pool all contributing perceptual factors into a single rating of basic audio quality. In order to improve understanding of the trade-offs between selected sound quality degradations that might arise in systems for the delivery of high quality multichannel audio, it was necessary to evaluate the influence of timbral and spatial fidelity changes on basic audio quality grades. The relationship between listener ratings of degraded multichannel audio quality on one timbral and two spatial fidelity scales was exploited to predict basic audio quality ratings of the same material using a regression model. It was found that timbral fidelity ratings dominated but that spatial fidelity predicted a substantial proportion of the basic audio quality.