关键词： 贝叶斯神经网络   支持向量回归机   机器学习算法   蒸散发数据插补  

摘要： 蒸散发是陆面生态过程的重要参数，也是流域水循环研究和水资源管理中最重要的因素之一。涡度相关技术的发展，为直接获取生态系统尺度的蒸散发数据提供了手段。但由于设备故障、天气恶劣以及数据质量控制等原因，基于涡度相关观测技术获取的蒸散发数据存在不同程度的缺失。如何利用科学合理的方法插补这些缺失的观测数据，在通量研究中具有重要的作用。\n 传统的蒸散发数据插补方法包括模型方法、查找表方法，为获取时间连续，质量保障，具有生态学机制的蒸散发数据提供了支撑。由于生态系统蒸散发和环境变量之间存在复杂的非线性关系，常规的统计方法难以实现缺失蒸散数据的插补。机器学习方法为揭示环境变量和生态系统蒸散发的非线性关系提供了可能。本研究试图从数据挖掘的角度开展蒸散发数据插补方法研究，以中国生态系统研究网络千烟洲森林生态系统观测站为例，结合计算机科学和生态学知识，采用贝叶斯神经网络、支持向量回归机、Factorization Machies和TimeFM等机器学习的方法来对蒸散发数据数值进行插补，从而探索一种新的数据插补方法实现生态系统蒸散发数据的高效插补。以下是论文的主要贡献:\n 1.将贝叶斯神经网络、支持向量回归机和Factorization Machies算法应用于没有包含时间属性的蒸散发数据的插补上面，并提出一种算法融合方法将三种算法进行融合，实现多种算法融合预测来提高预测准确度。\n 2.基于环境因子之间是相互影响且影响效果具有时序性的假设前提，本文提出并设计一个动态预测模型TimeFM。对比其它算法模型的预测结果，TimeFM能够很好地对环境因子数据进行建模并能准确地预测包含时间属性的蒸散发数据，平均均方误差相比对比算法下降15％～33％。

关键词： 归纳学习   解释学习   遗传算法   知识引导系数   知识发现  

摘要： 机器学习是解决故障诊断,数据挖掘,模式识别等众多领域中分类器建模的一项核心技术。在实际问题当中,普遍存在的一种问题是：所研究对象的先验领域知识不完备或者缺少能够完全覆盖各种模式的各类训练样本。在这种情况下基于传统的归纳学习策略或者解释学习策略所构建的分类器,其泛化能力较低,表现为在处理不同批次样本时,误判率较高,使其不能满足解决实际问题的需求。因此,在以解决此类问题为目标所构建的机器学习模型当中,须引入知识发现和知识修正机制,通过不断地发现新知识和修正错误知识,提高分类器的泛化能力。 针对上述问题,本文研究了一种基于解释-归纳策略的机器学习方法。利用所研究对象的先验知识,基于解释策略,对先验知识库进行建模,并利用遗传算法,通过先验知识引导下的种群初始化,变异函数的构造,基于特征评价准则的适应度函数的构造,建立知识发现体系模型,解决知识发现问题。最后,基于归纳策略,构建决策树分类器,并修改知识引导系数和知识属性参数,对错误知识进行修正,从而解决知识修正问题。 本文以CH5M1A型汽车变速箱为实验对象,利用现场所采集数据,采用本文的方法对变速箱各个档位的原始知识库进行进行知识发现和知识修正,并重新构建各个档位的决策树分类器,最后利用不同批次的测试样本进行测试,通过实验结果来看利用该方法可以有效解决分类器得知识发现和知识修正问题。

关键词： multicollinearity   factor analysis   statistics   regression   genetic programming   artificial neural network   ANN   machine learning   principal component analysis   PCA  

摘要： Multicollinearity occurs in a dataset due to correlation between the predictors. Models derived from such data without a check on multicollinearity may lead to erroneous system analysis. This problem can be eliminated by the selection of appropriate predictors from the dataset. Variable reduction methods like B2, B4, VIF, KIF and factor analysis (FA) can be used to overcome this problem. Such methods are useful particularly when used in conjunction with modelling methods that do not automate variable selection, such as artificial neural network (ANN) and fuzzy logic. The literature reveals that the current problem is aptly described in the field of statistics but is paid little attention in the field of machine learning. In this paper, multicollinearity is presented involving the estimation of fat content inside the body. Commonly used statistical methods such as stepwise regression, radial basis function partial least squares, partial robust M-regression, ridge regression and principal component regression are applied to this problem. The machine learning methods FA-ANN and genetic programming are also applied. The results are discussed with the interpretation and comparison of the modelling methods summarised in order to guide users on the proper techniques for tackling the multicollinearity problem.

关键词： 介词短语   依存句法分析   机器学习方法   介词短语边界  

摘要： 句法分析是自然语言处理研究过程中的重要技术之一。近几年依存句法分析的研究也越来越被关注。但相对于英文的依存句法分析研究成果而言汉语的依存句法分析还处于初级阶段。为此本文以现有的汉语依存句法分析研究成果为指导，采用基于决策式句法分析模型和基于分类器的训练算法相结合的依存句法分析方法，结合汉语固有的语法特点，针对汉语句子中含有介词短语的句法分析问题进行了研究和探索。 介词短语是一种复杂的短语结构，在汉语句子中占有很大的比例，汉语介词短语的研究主要是解决边界问题，正确识别介词短语对于汉语句子的依存关系分析具有重要的意义。本文首先对一些常见的依存句法分析方法进行了讲述，其中详细研究了Arc-eager算法和RPP算法,然后针对Arc-eager算法和RPP算法在分析过程中出现的错误问题，本文对RPP算法的介词短语边界识别策略做了相关的改进。由于Arc-eager算法只能处理邻近词之间的依存关系，所以Arc-eager算法在处理长距离右依存时会出现错误，RPP算法虽然对Arc-eager算法做了相关的改进，但是RPP算法在处理含有动词的介词短语时往往会出现错误分析的问题。为此，本文在对汉语句子进行依存关系分析时，根据汉语介词短语的固有特点，借鉴RPP算法的右边界后续词特征对介词短语进行边界识别，同时对RPP算法进行了一定的改进。经过改进后的介词短语依存关系分析算法，不但可以解决介词短语长距离分析时出现的过早依存问题，而且还可以处理含有动词的介词短语边界识别问题。 通过采用哈工大的依存关系语料库作为实验数据和基于支持向量机的句法分析器作为实验工具对本文提出的汉语句法分析策略进行了全面的测试和评估。实验结果表明本文提出的策略对汉语句子中含有介词短语的句法分析是有效的，而且显著的提高了分析的准确率。

摘要： Recently, the greatest statistical computational challenge in genetic epidemiology is to identify and characterize the genes that interact with other genes and environment factors that bring the effect on complex multifactorial disease. These gene-gene interactions are also denoted as epitasis in which this phenomenon cannot be solved by traditional statistical method due to the high dimensionality of the data and the occurrence of multiple polymorphism. Hence, there are several machine learning methods to solve such problems by identifying such susceptibility gene which are neural networks (NNs), support vector machine (SVM), and random forests (RFs) in such common and multifactorial disease. This paper gives an overview on machine learning methods, describing the methodology of each machine learning methods and its application in detecting gene-gene and gene-environment interactions. Lastly, this paper discussed each machine learning method and presents the strengths and weaknesses of each machine learning method in detecting gene-gene interactions in complex human disease.

关键词： Computer science   Biomedical Data   High-Dimensional   Imbalanced   Machine Learning  

摘要： Learning from high dimensional biomedical data attracts lots of attention recently. High dimensional biomedical data often suffer from the curse of dimensionality and have imbalanced class distributions. Both of these features of biomedical data, high dimensionality and imbalanced class distributions, are challenging for traditional machine learning methods and may affect the model performance. In this thesis, I focus on developing learning methods for the high-dimensional imbalanced biomedical data. In the first part, a sparse canonical correlation analysis (CCA) method is presented. The penalty terms is used to control the sparsity of the projection matrices of CCA. The sparse CCA method is then applied to find patterns among biomedical data sets and labels, or to find patterns among different data sources. In the second part, I discuss several learning problems for imbalanced biomedical data. Note that traditional learning systems are often biased when the biomedical data are imbalanced. Therefore, traditional evaluations such as accuracy may be inappropriate for such cases. I then discuss several alternative evaluation criteria to evaluate the learning performance. For imbalanced binary classification problems, I use the undersampling based classifiers ensemble (UEM) strategy to obtain accurate models for both classes of samples. A small sphere and large margin (SSLM) approach is also presented to detect rare abnormal samples from a large number of subjects. In addition, I apply multiple feature selection and clustering methods to deal with high-dimensional data and data with highly correlated features. Experiments on high-dimensional imbalanced biomedical data are presented which illustrate the effectiveness and efficiency of my methods.

关键词： trading strategies   high frequency trading   machine learning   clustering  

摘要： Not long ago securities were traded by human traders in face-to-face markets. The ecosystem of an open outcry market was well-known, visible to a human eye, and rigidly prescribed. Now trading is increasingly done in anonymous electronic markets where traders do not have designated functions or mandatory roles. In fact, the traders themselves have been replaced by algorithms (machines) operating with little or no human oversight. While the process of electronic trading is not visible to a human eye, machine-learning methods have been developed to recognize persistent patterns in the data. In this study, we develop a dynamic machine-learning method that designates traders in an anonymous electronic market into five persistent categories: high frequency traders, market makers, opportunistic traders, fundamental traders, and small traders. Our method extends a plaid clustering technique with a smoothing framework that filters out transient patterns. The method is fast, robust, and suitable for a discovering trading ecosystems in a large number of electronic markets.

关键词： 贷款违约预测   非平衡数据   随机森林   并行计算  

摘要： 如何在发放贷款前有效的评价和识别借款人潜在违约风险,计算借款人的违约概率,是现代金融机构信用风险管理的基础和重要环节,也是数量经济学、金融学等领域中的研究热点问题。 现有的贷款违约数据大部分都是非平衡的,以往的研究并未足够注意这一特征也就重视这一问题的深入研究,本文主要研究如何借助非平衡数据分类的思想对银行等金融机构的历史贷款数据进行分析,预测贷款违约的可能性。对于非平衡数据问题,采用基于数据平衡的方法构造随机森林；针对数据较大的问题,采用具有并行特性的随机森林算法。基于上述研究,本文提出了一种改进的带权重的并行平衡随机森林算法(WPBRF)。WPBRF算法在构造随机森林的每个决策树的同时利用OOB数据估计该决策树的预测性能,并据此赋予每个决策树不同的权重；此外,WPBRF算法利用了随机森林算法的可并行计算的特点,减少了单个决策树的训练时间。 实验结果表明,WPBRF在准确率和平衡准确率等方面超过了SVM、KNN、C4.5等常见分类算法和随机森林算法。此外,利用随机森林的并行性的WPBRF算法大幅降低了算法的学习时间,提高了算法的执行效率。

关键词： Overhead distribution system   Distribution reliability   Weather & animal-related outages   Ensemble learning methods  

摘要： Because a majority of day-to-day activities rely on electricity, it plays an important role in daily life. In this digital world, most of the people's life depends on electricity. Without electricity, the flip of a switch would no longer produce instant light, television or refrigerators would be nonexistent, and hundreds of conveniences often taken for granted would be impossible. Electricity has become a basic necessity, and so any interruption in service due to disturbances in power lines causes a great inconvenience to customers. Customers and utility commissions expect a high level of reliability. Power distribution systems are geographically dispersed and exposure to environment makes them highly vulnerable part of power systems with respect to failures and interruption of service to customers. Following the restructuring and increased competition in the electric utility industry, distribution system reliability has acquired larger significance. Better understanding of causes and consequences of distribution interruptions is helpful in maintaining distribution systems, designing reliable systems, installing protection devices, and environmental issues. Various events, such as equipment failure, animal activity, tree fall, wind, and lightning, can negatively affect power distribution systems. Weather is one of the primary causes affecting distribution system reliability. Unfortunately, as weather-related outages are highly random, predicting their occurrence is an arduous task. To study the impact of weather on overhead distribution system several models, such as linear and exponential regression models, neural network model, and ensemble methods are presented in this dissertation. The models were extended to study the impact of animal activity on outages in overhead distribution system. Outage, lightning, and weather data for four different cities in Kansas of various sizes from 2005 to 2011 were provided by Westar Energy, Topeka, and state climate office at K

关键词： Applied statistics   Logistic regression   New product development  

摘要： Critical success factors in new product development (NPD) in the Brazilian small and medium enterprises (SMEs) are identified and analyzed. Critical success factors are best practices that can be used to improve NPD management and performance in a company. However, the traditional method for identifying these factors is survey methods. Subsequently, the collected data are reduced through traditional multivariate analysis. The objective of this work is to develop a logistic regression model for predicting the success or failure of the new product development. This model allows for an evaluation and prioritization of resource commitments. The results will be helpful for guiding management actions, as one way to improve NPD performance in those industries. © Universidad Alberto Hurtado, Facultad de Economía y Negocios.