奇异数据筛选法在玉米籽粒蛋白质近红外光谱检测中的应用

相对于传统生化测定方法,基于近红外光谱(Near infrared spectroscopy,NIRS)玉米籽粒蛋白质含量检测是一种快速、非破坏、且适用于多组分同时检测的新方法。但在建模过程中,由于奇异数据(异常值)的存在会影响近红外光谱模型的预测精度和稳定性,我们采用奇异数据筛选法剔除了玉米籽粒近红外光谱中的奇异数据并建立了玉米籽粒蛋白质含量的偏最小二乘支持向量机(Least squares support vector machine,LS-SVM)模型。本文分别采用杠杆值法(Leverage)、半数重采样法(Resampling by Half-Mean,RHM)和蒙特卡洛采样法(Monte-Carlo Sampling,MCS)剔除了玉米籽粒蛋白质光谱数据中的奇异数据并对模型结果进行比较。在剔除奇异数据的基础上,采用偏最小二乘回归法(Partial least squares regression,PLSR)提取主成分,并基于小生境蚁群算法(Niche ant colony algorithm,NACA)优化偏最小二乘支持向量机(LS-SVM)模型参数(γ和σ2),建立基于LS-SVM的玉米籽粒蛋白质定量分析模型。结果表明,采用3种奇异数据筛选法剔除奇异数据后所建LS-SVM模型的预测结果都优于采用原光谱数据所建模型,相比较而言,蒙特卡洛采样法为基于近红外光谱检测玉米籽粒蛋白质的最佳奇异数据筛选法。     

采用最小二乘支持向量机的青霉素发酵过程建模研究

生化过程通常是严重非线性和时变的复杂动态系统,而且重要过程参数缺少在线测量仪表,对其建立机理模型往往非常耗时和困难。采用最小二乘支持向量机(LS_SVM)并以Pensim仿真平台为例对青霉素发酵这一典型生化过程进行建模研究。给出了LS_SVM参数的调整策略和分析结果,建立了青霉素产物浓度、菌体浓度和底物浓度等重要过程变量的在线预报模型。仿真结果表明用LS_SVM建立的在线预报模型拟合误差小,推广性能好,可以作为发酵过程的进一步控制和优化的参考依据。     

两种预测模型在地下水动态中的比较与应用

预测陕西洛惠渠灌区地下水动态变化情况,在综合分析了各种地下水动态研究方法的基础上,提出了基于支持向量机和改进的BP神经网络模型的灌区地下水动态预测方法,并在MATLAB中编制了相应的计算机程序,建立了相应的地下水动态预测模型。以灌区多年实例数据为学习样本和测试样本,比较了两种模型的地下水动态预测优劣性。研究表明,支持向量机模型和BP网络模型在样本训练学习过程中都具较高的模拟精度,而在样本学习阶段,支持向量机的预测精度明显优于BP网络,可以很好的描述地下水动态复杂的耦合关系。支持向量机方法切实可行,更加适合大型灌区地下水动态预测,是对传统地下水动态研究方法的补充与完善。     

基于近红外光谱实时监测乙醇发酵过程多组分参数

生物量、葡萄糖浓度和乙醇浓度是乙醇发酵过程的重要参数,传统的方法通常对发酵液取样作离线测量,不仅需要采用多种仪器进行测试分析,而且耗时耗力,成为实时过程调控和优化的障碍。文中针对这些重要过程参数提出了一个基于近红外光谱技术的原位实时检测方法。通过采用浸入式近红外光谱仪对发酵溶液进行原位测量,基于多输出最小二乘支持向量机回归(MLS-SVR)方法建立了利用近红外光谱同时分析葡萄糖浓度、生物量和乙醇浓度的多输出预测模型。实验结果表明,该方法能实时准确地检测乙醇发酵过程中的葡萄糖浓度、生物量和乙醇浓度,而且相对于现有的偏最小二乘法(PLS)分别对各组分建模和预测,能明显提高测量准确性和可靠性。     

基于PSO-LSSVR模型的土壤肥力评价

通过引入粒子群算法(PSO)和最小二乘支持向量机(LSSVR),提出基于PSO-LSSVR的土壤肥力评价模型。选取有机质、全氮速效磷、速效钾、阳离子交换量、酸碱度、容重、黏粒、水稳性团聚体和分散率等10种评价指标,以吉林省黑地为例,建立土壤肥力评价模型。同时与物元可拓法、普通SVM模型的评价结果进行比较;3种方法的多数样本评价结果基本一致,对于样本2、样本13,PSO-LSSVR模型分别定为Ⅳ级、Ⅲ级,符合实际情况;表明PSO-LSSVR是一种适用且能准确反映土壤特性的土壤肥力评价模型。     

基于SVR算法的林地土壤氮含量高光谱测定

提出了一种利用高光谱技术进行杉木林土壤全氮测定的新方法。以FieldSpec®3地物光谱仪采集杉木林土壤148份, 随机分成校正集(100份)和检验集(48份)。以不同方法实现了土壤光谱的预处理, 并采用偏最小二乘回归算法(PLS)建立土壤氮含量估测模型对其进行比较分析, 发现小波除噪结合多元散射校正能最有效地消除原始光谱的噪声与背景信息, 此时PLS模型校正集与预测集R²分别为0.891与0.885。为进一步优化模型, 对经小波除噪结合多元散射校正处理后的光谱采用主成分分析法(PCA)降维, 以前4个主成份为输入变量, 采用小二乘支持向量机回归算法(LS-SVR)建立了土壤氮含量估测模型, 其校正集与预测集R²分别提高至0.921与0.917, 具有比PLS算法更高的精度。结果表明:以高光谱技术进行林地土壤氮含量快速监测是可行的, 其中小波去噪结合多元散射校正系光谱预处理的优选方法, 而LS-SVR则是建模的优选方法。     

支持向量机与神经网络的关系研究

支持向量机是一种基于统计学习理论的新颖的机器学习方法,由于其出色的学习性能,该技术已成为当前国际机器学习界的研究热点,该方法已经广泛用于解决分类和回归问题．本文将结构风险函数应用于径向基函数网络学习中,同时讨论了支持向量回归模型和径向基函数网络之间的关系．仿真实例表明所给算法提高了径向基函数网络的泛化性能．     

三种回归分析方法在Hyperion影像LAI反演中的比较

借助GPS进行地面精确定位,利用LAI-2000冠层分析仅在攸县黄丰桥林场开展130个样地(60m×60m)的叶面积指数(Leaf Area Index,LAI)测量.采用FLAASH模块对Hyperion数据进行大气校正并与地面同步冠层观测数据进行拟合,通过研究地面实测LAI与Hyperion影像波段及其衍生的系列植被指数(NDVI、RVI等)的相关性,筛选出估算叶面积指数的植被指数因子.应用曲线估计、逐步回归及偏最小二乘三种回归分析技术分别建立叶面积指数的最优估算模型.结果表明:参与建模的因子中,比值植被指数(RVI)与LAI的相关性最大,敏感性最高,其次是SARVI0.1,NDVI705,NDVI,SARVI0.1,SARVI0.25;曲线估计、逐步回归分析和偏最小二乘回归三种分析方法所建的6个回归模型中,偏最小二乘回归的拟合效果最好,预测值与实测值的决定系数R2为0.84、曲线估计的拟合效果最低,预测值与实测值的决定系数R2为0.64;建模精度分析表明,选用5-6个自变量因子进行LAI建模是可靠的,以6个植被因子建立的偏最小二乘回归模型预测精度最高.     

基于近红外光谱的冬小麦籽粒蛋白质含量检测

冬小麦籽粒蛋白质含量(GPC)是评价冬小麦品质的主要指标,为了研究不同建模方法对GPC检测的影响,本研究对冬小麦籽粒的近红外原始光谱进行S-G平滑、基线校正和多元散射校正等预处理,利用连续投影算法(SPA)提取冬小麦GPC的重要光谱波段,并结合偏最小二乘回归(PLSR)、主成分回归(PCR)、支持向量机(SVM)和多元线性回归(MLR)建立GPC的光谱预测模型,并综合比较模型的适用性。结果表明:经过SPA提取的特征波段为1801、1010、1109、2284、2219、2239、871、1361、1925、1849和1456 nm;模型评价方面,利用特征波段建立的SVM模型效果较好,其中校正均方根误差(RMSEC)和R2分别为0.2481和0.9760,验证均方根误差(RMSEP)和R2分别为0.3587和0.9581。研究表明,SPA+SVM预测模型在一定程度上能够实现冬小麦籽粒蛋白质的快速、无损检测。     

食用调和油中花生油含量的近红外光谱分析

采用偏最小二乘法(PLS)等方法建立了食用调和油中花生油含量定量分析的近红外光谱定标模型。采集食用调和油样品在4 000 cm-1~10 000 cm-1范围内的近红外漫反射光谱,光谱经一阶导数处理后,采用偏最小二乘法建立样品中花生油含量的定标模型,并用Leave-one-out内部交叉验证法对模型进行验证。模型相关系数为0.99961,校正均方根RMSEC为0.830%。比较不同光谱预处理方法对定标模型的影响,结果表明一阶导数Corr.coeff最好。采用不同的化学计量学方法建立的定标模型中以偏最小二乘回归法最理想。     

Soft-sensing method based on FDLS-SVM in marine alkaline protease fermentation process

Abstract

To overcome the problem that soft-sensing model cannot be updated with the bioprocess changes, this article proposed a soft-sensing modeling method which combined fuzzy c-means clustering (FCM) algorithm with least squares support vector machine theory (LS-SVM). FCM is used for separating a whole training data set into several clusters with different centers, each subset is trained by LS-SVM and sub-models are developed to fit different hierarchical property of the process. The new sample data that bring new operation information is introduced in the model, and the fuzzy membership function of the sample to each clustering is first calculated by the FCM algorithm. Then, a corresponding LS-SVM sub-model of the clustering with the largest fuzzy membership function is used for performing dynamic learning so that the model can update online. The proposed method is applied to predict the key biological parameters in the marine alkaline protease MP process. The simulation result indicates that the soft-sensing modeling method increases the model’s adaptive abilities in various operation conditions and can improve its generalization ability.     

Predictive modeling of human operator cognitive state via sparse and robust support vector machines

The accurate prediction of the temporal variations in human operator cognitive state (HCS) is of great practical importance in many real-world safety-critical situations. However, since the relationship between the HCS and electrophysiological responses of the operator is basically unknown, complicated and uncertain, only data-based modeling method can be employed. This paper is aimed at constructing a data-driven computationally intelligent model, based on multiple psychophysiological and performance measures, to accurately estimate the HCS in the context of a safety-critical human–machine system. The advanced least squares support vector machines (LS-SVM), whose parameters are optimized by grid search and cross-validation techniques, are adopted for the purpose of predictive modeling of the HCS. The sparse and weighted LS-SVM (WLS-SVM) were proposed by Suykens et al. to overcome the deficiency of the standard LS-SVM in lacking sparseness and robustness. This paper adopted those two improved LS-SVM algorithms to model the HCS based solely on a set of physiological and operator performance data. The results showed that the sparse LS-SVM can obtain HCS models with sparseness with almost no loss of modeling accuracy, while the WLS-SVM leads to models which are robust in case of noisy training data. Both intelligent system modeling approaches are shown to be capable of capturing the temporal fluctuation trends of the HCS because of their superior generalization performance.     

Statistical analysis of big data: an approach based on support vector machines for classification and regression problems

A new type of learning algorithms with the supervisor for estimating multidimensional functions is considered. These methods based on Support Vector Machines are widely used due to their ability to deal with high-dimensional and large datasets, and their flexibility in modeling diverse sources of data. Support vector machines and related kernel methods are extremely good at solving prediction problems in computational biology. A background about statistical learning theory and kernel feature spaces is given including practical and algorithmic considerations.     

Predicting breast cancer using an expression values weighted clinical classifier

Background

Clinical data, such as patient history, laboratory analysis, ultrasound parameters-which are the basis of day-to-day clinical decision support-are often used to guide the clinical management of cancer in the presence of microarray data. Several data fusion techniques are available to integrate genomics or proteomics data, but only a few studies have created a single prediction model using both gene expression and clinical data. These studies often remain inconclusive regarding an obtained improvement in prediction performance. To improve clinical management, these data should be fully exploited. This requires efficient algorithms to integrate these data sets and design a final classifier.LS-SVM classifiers and generalized eigenvalue/singular value decompositions are successfully used in many bioinformatics applications for prediction tasks. While bringing up the benefits of these two techniques, we propose a machine learning approach, a weighted LS-SVM classifier to integrate two data sources: microarray and clinical parameters.

Results

We compared and evaluated the proposed methods on five breast cancer case studies. Compared to LS-SVM classifier on individual data sets, generalized eigenvalue decomposition (GEVD) and kernel GEVD, the proposed weighted LS-SVM classifier offers good prediction performance, in terms of test area under ROC Curve (AUC), on all breast cancer case studies.

Conclusions

Thus a clinical classifier weighted with microarray data set results in significantly improved diagnosis, prognosis and prediction responses to therapy. The proposed model has been shown as a promising mathematical framework in both data fusion and non-linear classification problems.     

中亚热带人工针叶林生态系统碳通量拆分差异分析

涡度通量观测可直接获取陆地生态系统与大气之间CO2净交换量(NEE),但深入认识碳循环过程和校验生态系统模型需要不同时间尺度总初级生产力(GPP)和生态系统呼吸(Re)等碳通量数据。利用中国陆地生态系统通量观测与研究网络(ChinaFLUX)中亚热带人工针叶林生态系统2003—2009年的涡度通量和气象观测数据,分析了两种NEE拆分方法对不同时间尺度GPP和Re评估的影响,结果表明:(1)两种拆分方法得到的生态系统碳通量组分(GPP和Re)的季节动态变化一致,都在生长季7、8月份达到峰值;(2)非线性回归模型拆分得到的全年Re和GPP相较于光响应曲线模型分别高出2%—28.6%和1.6%—23%,最大高出317.6 gC·m-2·a-1(2006年),逐月最大差值主要发生在8、9月份;(3)不同时间尺度上,两种方法拆分得到的GPP和Re之间差值的环境响应因子不同。在广泛采用非线性回归模型进行拆分时,如果当月光合有效辐射接近到905mol·m-2·月-1,月平均空气饱和水汽压差接近1.18 kPa时,需要考虑使用光响应曲线模型拆分该月通量,结合两种拆分方法以减小全年的误差。     

Directed molecular evolution by machine learning and the influence of nonlinear interactions

Alternative search strategies for the directed evolution of proteins are presented and compared with each other. In particular, two different machine learning strategies based on partial least-squares regression are developed: the first contains only linear terms that represent a given residue's independent contribution to fitness, the second contains additional nonlinear terms to account for potential epistatic coupling between residues. The nonlinear modeling strategy is further divided into two types, one that contains all possible nonlinear terms and another that makes use of a genetic algorithm to select a subset of important interaction terms. The performance of each modeling type as a function of training set size is analysed. Simulated molecular evolution on a synthetic protein landscape shows the use of machine learning techniques to guide library design can be a powerful addition to library generation methods such as DNA shuffling.     

Statistical learning techniques applied to epidemiology: a simulated case-control comparison study with logistic regression

Background  

When investigating covariate interactions and group associations with standard regression analyses, the relationship between the response variable and exposure may be difficult to characterize. When the relationship is nonlinear, linear modeling techniques do not capture the nonlinear information content. Statistical learning (SL) techniques with kernels are capable of addressing nonlinear problems without making parametric assumptions. However, these techniques do not produce findings relevant for epidemiologic interpretations. A simulated case-control study was used to contrast the information embedding characteristics and separation boundaries produced by a specific SL technique with logistic regression (LR) modeling representing a parametric approach. The SL technique was comprised of a kernel mapping in combination with a perceptron neural network. Because the LR model has an important epidemiologic interpretation, the SL method was modified to produce the analogous interpretation and generate odds ratios for comparison.     

旱改水型农田整治对土壤碳排放的短期影响

灌溉农业可提升粮食生产潜力,已成为全球农业重要的发展方向,但此类土地利用转换势必影响旱作农田土壤的稳定性,尤其是碳循环。然而,旱改水整治过程中土壤碳通量变化及其与环境因子间的互馈机制尚不清楚。为此,采用大田模拟实验,连续7 d监测土壤碳通量变化,评估旱改水整治对土壤碳库组成及环境驱动的短期效应。结果表明：①旱地、水田的土壤碳通量和温度均呈昼高夜低的单峰型曲线,且碳通量与温度峰值出现于每日13：00前后,但水田土壤碳通量稍高于旱地。②旱改水后短期内土壤可溶性有机碳（DOC）、微生物量碳（MBC）、易氧化有机碳（EOC）、惰性有机碳（ROC）、总有机碳（TOC）和土壤碳库管理指数均呈减少趋势,其中土壤微生物量碳、易氧化有机碳降幅分别达28.55%、29.09%。③土壤含水量、微生物OTU数、碳库含量是影响碳通量速率变化的关键因子（P<0.05）,土壤温度、理化性状是制约土壤碳库的主控因子（P<0.05）。农业活动是重要的碳源之一,深入研究大范围旱改水诱发的碳排放问题可为低碳农业、气候减缓及其应对策略制定提供科学依据。