Technical Note: Bias and the Quantification of Stability

Research on bias in machine learning algorithms has generally been concerned with the impact of bias on predictive accuracy. We believe that there are other factors that should also play a role in the evaluation of bias. One such factor is the stability of the algorithm; in other words, the repeatability of the results. If we obtain two sets of data from the same phenomenon, with the same underlying probability distribution, then we would like our learning algorithm to induce approximately the same concepts from both sets of data. This paper introduces a method for quantifying stability, based on a measure of the agreement between concepts. We also discuss the relationships among stability, predictive accuracy, and bias.     

Technical Note: Some Properties of Splitting Criteria

Machine Learning - Various criteria have been proposed for deciding which split is best at a given node of a binary classification tree. Consider the question: given a goodness-of-split criterion...     

Uncertain reasoning using the Dempster-Shafer method: an application in forecasting and marketing management

Abstract: The intended purpose of this article is twofold: to study techniques for uncertainty management in expert systems, particularly the Dempster-Shafer theory of belief; and to use this method in the construction of an expert system for the field of forecasting and marketing management. Compared with the probabilistic approach, which can only deal with singleton possibilities, the Dempster-Shafer approach proves to be superior because it provides the ability to deal with sets of possibilities. Since market analysis and forecasting always have a strong element of uncertainty associated with them, and since, in general, we consider a combination of several different forecasting techniques for planning our marketing strategies, the Dempster-Shafer approach is particularly suitable. Here we present a short introduction to this theory, briefly discuss the domain of application (selection of forecasting techniques for marketing planning), discuss the interesting components of our expert system, and analyze our experiences in applying the theory to this domain.     

Technical Note: The use of mirrors during an assembly task: a study of ergonomics and productivity

Industrial assembly tasks often require awkward, sustained neck and/or shoulder postures that can lead to increased musculoskeletal discomfort and reduced productivity. The aim of this study was to investigate the effects of mirror and periscope visual aids as ergonomic interventions designed to eliminate awkward postures of the cervicobrachial region during assembly tasks. Participants simulated a simple assembly task by using a cordless screwdriver to drive screws into a pre-tapped aluminium block. Trials of 15 min were run for each of four distinct assembly workstation configurations: industry standard (in-line screwdriver, work at elbow height, no visual aid); pistol grip (pistol grip screwdriver, work at shoulder height, no visual aid); mirror (in-line screwdriver, work at elbow height, single mirror visual aid); and periscope (in-line screwdriver, work at elbow height, two-mirror visual aid system). Muscular activity, discomfort, body posture, productivity and operator subjective assessment were recorded to determine the effects of the visual aid interventions. The results show that when comparing the interventions to the industry standard condition, there was a 45% reduction in average cervical erector spinae activity, a 90% reduction in average neck flexion angle and a 72% reduction in neck discomfort with the interventions. When comparing these interventions to the pistol grip condition there was an 80% reduction in activity of the dominant side deltoid, a 92% reduction in shoulder flexion angle and an 81% decrease in shoulder discomfort with the interventions. Productivity was greatest in the industry standard configuration followed by the pistol grip (9% lower), the periscope (13% lower) and the mirror (23% lower) configurations. A follow-up study that compared the productivity of the periscope configuration with that of the industry standard configuration showed that within a 4-h work period this productivity differential decreased by over 33%.     

Comparing and classifying one-dimensional spatial patterns: an application to laser altimeter profiles

Numerical analyses of remotely sensed data may valuably contribute to an understanding of the vegetation/land surface interface by pointing out at which scales a given variable displays a high level of spatial variability. Thus, there is a need of methods aimed at classifying many one-dimensional signals, such as airborne laser profiles, on the basis of their spatial structure. The present paper proposes a theoretical framework ensuring a consistent combination of a multi-scale pattern characterization, based on the Haar wavelet variance (also called in ecology Two Terms Local Variance, TTLV), with two multivariate techniques such as principal components analysis (PCA) and hierarchical cluster analysis. We illustrate our approach by comparing and classifying 257 laser profiles, with a length of 64 measurements (448 m), that were collected by the BRGM in French Guiana over three main landscape units with distinct geomorphological and ecological characteristics. We calculate for each profile a scalogram that summarized the multi-scale pattern and analyze the structural variability of profiles via a typology and a classification of one-dimensional patterns. More than 80% of the variability between spatial patterns of laser profiles has been summarized by two PCA axes, while four classes of spatial patterns were identified by cluster analysis. Each landscape unit was associated with one or two dominant classes of spatial patterns. These results confirmed the ability of the method to extract landscape scaling properties from complex and large sets of remotely sensed data.     

Estimation and forecasting with logarithmic autoregressive conditional duration models: A comparative study with an application

This paper presents a semi-parametric method of parameter estimation for the class of logarithmic ACD (Log-ACD) models using the theory of estimating functions (EF). A number of theoretical results related to the corresponding EF estimators are derived. A simulation study is conducted to compare the performance of the proposed EF estimates with corresponding ML (maximum likelihood) and QML (quasi maximum likelihood) estimates. It is argued that the EF estimates are relatively easier to evaluate and have sampling properties comparable with those of ML and QML methods. Furthermore, the suggested EF estimates can be obtained without any knowledge of the distribution of errors is known. We apply all these suggested methodology for a real financial duration dataset. Our results show that Log-ACD (1, 1) fits the data well giving relatively smaller variation in forecast errors than in Linear ACD (1, 1) regardless of the method of estimation. In addition, the Diebold–Mariano (DM) and superior predictive ability (SPA) tests have been applied to confirm the performance of the suggested methodology. It is shown that the new method is slightly better than traditional methods in practice in terms of computation; however, there is no significant difference in forecasting ability for all models and methods.     

TRAPPY: a truthfulness and reliability aware application placement policy in fog computing

The Journal of Supercomputing - Fog computing facilitates the satisfaction of Internet of things (IoT) users by running time-sensitive offloaded applications. Heterogeneity, dynamism, and the...     

Comparison of forecasting methods with an application to predicting excess equity premium

This paper reviews various forecast methods including combination using theoretically optimal weights and those under model selection approaches. In addition, we suggest two modified simple averaging forecast combination methods—a mean corrected and a mean and scale corrected method. We conclude that due to the fact that real data is usually subject to structural breaks, rolling forecasting scheme has a better performance than fixed window and continuously updating scheme. In addition, methods that use less information appear to perform better than methods using all the sample information about the covariance structure of the available forecasts. The mean and scale corrected simple average approach yield smaller mean squared forecast error than the three widely used regression approaches suggested by Granger and Ramanathan [11].     

MAYA中的建模及贴图技术分析

本文详细分析了Maya中的建模技术和贴图技术,并提出需要注意的几点问题.     

基于OSG的实时烟雾模拟的研究与应用

为了模拟出具有实时性和真实感的烟雾效果,分析了传统烟雾的模拟方法,提出了一种基于OSG平台下的实时烟雾模拟方法。该方法采用流体力学纳维-斯托克斯方程组作为基本物理模型,通过添加漩涡约束力来达到真实性的要求。由于NS方程求解的复杂性和CPU计算的限制性,在方程解算上,提出一种基于GPU的采用半拉格朗日方法、雅克比迭代法等稳定的求解方法,并结合体绘制技术来实现实时烟雾的模拟。实验结果表明,该方法能快速模拟出实时、逼真的烟雾效果。     

PROBE spatial data modeling and query processing in an imagedatabase application

The PROBE research project has produced results in the areas of data modeling, spatial/temporal query processing, recursive query processing, and database system architecture for nontraditional application areas, many of which involve spatial data and data with complex structure. An overview of PROBE is provided, focusing on the facilities for dealing with spatial and temporal data. It is shown how the PROBE database system and simple application-specific object classes combine to efficiently support PROBE's spatial data model. It is also shown how an image-database application can be supported using PROBE's data model and spatial query processor. The current status of the PROBE project and future plans are discussed     

Estimation of winter wheat acreage via a combination of remotely sensed data and an optimized spatial sampling scheme

Timely and accurate estimates of crop areas are critical to enhancing agriculture management and ensuring national food security. This study aims to combine remote-sensing data and an optimized spatial sampling scheme to improve the accuracy of crop area estimates and decrease the cost of crop surveys at a regional scale. This study focuses on winter wheat in Mengcheng County in Anhui Province, China. Advanced Land Observing Satellite (ALOS) Advanced Visible light and Near Infrared Radiometer (AVNIR)-2, and Landsat5 Thematic Mapper (TM) images from 2009 and 2010, respectively, are used to extract the winter wheat area and distribution. Additionally, a spatial sampling scheme was optimized by combining remotely sensed data, geographical information system (GIS), Geostatistics, and traditional sampling methods. The experimental results demonstrate that the variability in the proportion of winter wheat acreage in one sampling unit (PWS) increases with increasing sampling unit size. The PWS coefficient of variation (CV) varies from 32.75 to 43.46% among the eight sampling unit sizes. The spatial correlation thresholds of PWS increase with increasing sampling unit size. For small sampling unit sizes (500 m × 500 m–2000 m × 2000 m), the relative error and CV of the population extrapolation for the optimized sample layout are obviously lower than those of the simple random sampling method. For larger sampling unit sizes (2500 m × 2500 m–4000 m × 4000 m), the sample size is obviously lower for the optimized sample layout compared with that of the simple random sampling method, but there are no differences in the relative errors or CVs. By combining remote-sensing data and the optimized spatial sampling scheme, this research can improve the accuracy of crop area estimation at a regional scale.     

Technical Note: Pseudo natural colour aerial imagery for urban and suburban mapping

Due to their near‐infrared data channel, digital airborne four‐channel imagers provide a potentially good discrimination between vegetation and human‐made materials, which is very useful in automated mapping. Due to their red, green and blue data channels, they also provide natural colour images, which are very useful in traditional (manual) mapping. In this paper, an algorithm is described which provides an approximation to the spectral capabilities of the four‐channel imagers by using a colour‐infrared aerial photo as input. The algorithm is tailored to urban/suburban surroundings, where the quality of the generated (pseudo) natural colour images are fully acceptable for manual mapping. This brings the combined availability of near‐infrared and (pseudo) natural colours within reach for mapping projects based on traditional photogrammetry, which is valuable since traditional analytical cameras still by far outnumber the relatively new family of digital airborne four‐channel imagers.     

物联网关键技术及典型军事应用

物联网技术能够实现物与物之间的互联和监控,因而必将引起军事领域的巨大变革,成为加速军队战斗力生成模式转变、推进军事变革的有力推手。文中描述了物联网的发展历程,介绍了物联网的技术框架和标准化进程以及物联网在军事上的典型应用,包括战场兵力呈现系统、战场物资保障综合系统和战术战法仿真验证系统。利用这些系统,可使指挥员实时感知战场的兵力分布和物资保障情况,并能通过获得的敌方兵力分布,来对不同的战术战法进行仿真,以验证其可行性和有效性。     

Technical Note: Crop stress detection using AVIRIS hyperspectral imagery and artificial neural networks

The objectives of this study were to compare the results of artificial neural network (ANN) and standard vegetation algorithm processing to distinguish nutrient stress from in-field controls, and determine whether nutrient stress might be distinguished from water stress in the same test field. The test site was the US Department of Agriculture's Variable Rate Application (VRAT) site, Shelton, Nebraska. The VRAT field was planted in corn with test plots that were differentially treated with nitrogen (N). The field contained four replicates, each with N treatments ranging from 0 kg ha^?1 to 200 kg ha^?1 in 50 kg ha^?1 increments. Low-altitude (3 m pixel) Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) hyperspectral imagery (224 bands) was collected over the site. Ground data were collected to support image interpretation. An ANN was applied to the AVIRIS image data for detection of crop and water stress. Known vegetation indices were used as a baseline for comparison against ANN-based stress detection. The resulting comparison found that ANN methods provided a heightened capability to separate stressed crops from in-field, non-stressed controls and was sensitive to differences in nutrient- and water-stressed field regions.     

Technical Note: Response of ratio vegetation indices to foot rot-infected citrus trees

Ratio vegetation indices were evaluated for differentiating foot rot-infected (Phytophthora parasitica Dast) from noninfected citrus trees (Citrus spp.). At two sites, field spectral measurements of infected and noninfected trees were obtained with a hand-held spectroradiometer. The following bands were assessed for developing the ratios: blue (B, 450–520?nm), green (G, 520–600?nm), red (R, 630–690?nm), near-infrared (NIR, 760–900?nm) and mid-infrared (MIR1, 1550–1750?nm; MIR2, 2080–2350?nm). For both sites, means of infected and noninfected trees were significantly different (t-test; p<0.05) for NIR/B, NIR/R and NIR/MIR1 ratios. Percentage difference values indicated the NIR/R ratio was most responsive to the disease. A relative index value was established for separating infected from noninfected trees. Findings demonstrated that ratio vegetation indices have high potential for distinguishing foot rot-infected from noninfected citrus trees.     

网络流量短期预测方法的研究与应用

给出了网络流量短期预测方法.该方法运用小波变换自适应时频局部化分析方法和改进的Mallat算法将网络流量分解到不同频带上,然后对各子频带上的小波分进行不同阈值的消噪处理,再对仍是非平稳过程的分量进行差分处理使其转化为平稳序列,最后对各平稳过程分量采用ARMA模型进行预测.实际流量分析表明该方法简便,且短期预测精度较高.