基于判别分析与ANN的药品铝塑包装泡罩XRF分析

目的针对案件现场常见的药品铝塑包装泡罩,为达到对其分类识别的目的,提出系列检验分析、数据处理方法。方法采用X射线荧光光谱法对45个药品铝塑包装泡罩样本所含元素进行检验并讨论分析。对检验结果进行无监督的系统聚类,利用离差平方和法计算欧氏距离进而将未知样本分为5类。结果将分类结果作为变量进行判别分析,选取累积方差百分比为97.8%的2个判别函数,其类内平方和与总平方和之比为0.015和0.394,具有较强的解释能力。绘制的样本判别分类图将5类样本类之间相互区分开来,样本总体判别正确率为95.6%。提取样本在判别函数上的判别得分构建了人工神经网络,最终分类正确率为97.8%。结论利用X射线荧光光谱法对药品铝塑包装泡罩进行检验,将元素种类及含量作为变量进行了分类,并构建了45个药品铝塑包装泡罩样本的人工神经网络分类模型,可借助该模型进一步实现对于案件现场未知类别的药品铝塑包装泡罩样本的分类识别。     

红外光谱结合主成分分析对纸质快递文件袋的分类研究

目的 建立一种快速无损的检验纸质快递文件袋的分析方法。方法 利用傅里叶变换红外光谱对63个纸质快递文件袋样品进行检验，分析样品的红外光谱吸收峰的峰位，结合主成分分析对光谱数据进行了降维处理并分类。利用费歇尔判别对快递文件袋的分类结果进行分析和验证。同时建立多层感知器神经网络和径向基函数神经网络2种分类模型，进行分析和验证。结果 63个纸质快递文件袋样品可被分成四大类，利用费歇尔分类模型进行验证，准确率为100%；多层感知器神经网络分类模型准确率为95.23%，径向基函数神经网络分类模型准确率为92.06%。通过比较发现，费歇尔判别可以实现对纸质快递文件袋更加有效地分类。结论 该方法简单快速，样品用量少且无损样品，可为快递文件袋类的物证鉴定提供科学依据。     

Application of an artificial neural network in chromatography—retention behavior prediction and pattern recognition

Layered feed-forward neural networks are powerful tools particularly suitable for the analysis of nonlinear multivariate data. In this paper, an artificial neural network using improved error back-propagation algorithm has been applied to solve problems in the field of chromatography. In this paper, an artificial neural network has been used in the following two applications: (1) To model retention behavior of 32 solutes in a methanol–tetrahydrofuran–water system and 49 solutes in methanol–acetonitrile–water system as a function of mobile phase compositions in high performance liquid chromatography. The correlation coefficients between the calculated and the experimental capacity factors were all larger than 0.98 for each solute in both the training set and the predicting set. The average deviation for all data points was 8.74% for the tetrahydrofuran-containing system and 7.33% for the acetonitrile-containing system. 2). To classify and predict two groups of different liver and bile diseases using bile acid data analyzed by reversed-phase high performance liquid chromatography (RP-HPLC). The first group includes three classes: healthy persons, choledocholithiasis patients and cholecystolithiasis patients; the total consistent rate of classification was 87%. The second group includes six classes: healthy persons, pancreas cancer patients, hepatoportal high pressure patients, cholelithiasis patients, cholangietic jaundice patients and hepatonecrosis patients; the total consistent rate of classification was 83%. It was shown that artificial neural network possesses considerable potential for retention prediction and pattern recognition based on chromatographic data.     

Application of supervised self-organizing maps in breast cancer diagnosis by total synchronous fluorescence spectroscopy

Data from total synchronous fluorescence spectroscopy (TSFS) measurements of normal and malignant breast tissue samples are introduced in supervised self-organizing maps, a type of artificial neural network (ANN), to obtain diagnosis. Three spectral regions in both TSFS patterns and first-derivative TSFS patterns exhibited clear differences between normal and malignant tissue groups, and intensities measured from these regions served as inputs to neural networks. Histology findings are used as the gold standard to train self-organizing maps in a supervised way. Diagnostic accuracy of this procedure is evaluated with sample test groups for two cases, when the neural network uses TSFS data and when the neural network uses data from first-derivative TSFS. In the first case diagnostic sensitivity of 87.1% and specificity of 91.7% are found, while in the second case sensitivity of 100% and specificity of 94.4% are achieved.     

Rule Extraction from Trained Artifical Neural Networks

The neural network rule extraction problem is aimed at obtaining rules from an arbitrarily trained artificial neural network. Recently there have been several approaches to rule extraction. Approaches to rule extraction implement a priori knowledge of data or rule requirements into neural networks before the rules are extracted. Although this may lead to a simplified final phase of acquitting the rules from particular type of neural networks, it limits the methodologies for general-purpose use. This article approaches the neural network rule extraction problem in its essential and general form. Preference is given to multilayer perceptron networks (MLP networks) due to their universal approximation capabilities. The article establishes general theoretical grounds for rule extraction from trained artificial neural networks and further focuses on the problem of crisp rule extraction. The problem of crisp rule extraction from trained MLP networks is first approached on theoretical level. Present ed theoretical results state conditions guaranteeing equivalence between classification by an MLP network and crisp logical formalism. Based on the theoretical results an algorithm for crisp rule extraction, independent of training strategy, is proposed. The rule extraction algorithm can be used even in cases where the theoretical conditions are not strictly satisfied; by offering an approximate classification. An introduced rule extraction algorithm is experimentally demonstrated.

     

Supervised pattern recognition to authenticate Italian extra virgin olive oil varieties

This work is the logical consecution of our previous investigation on the classification of “monocultivar” olive oils, in which 153 samples from the five cultivars Carboncella, Frantoio, Leccino, Moraiolo and Pendolino [harvested from 1997 to 1999, in the same geographical area (Sabina, Lazio)] were discriminated according to their variety, using Linear Discriminant Analysis (LDA) and Back-propagation Artificial Neural Network (BP-ANN).This study has been now extended to include 50 new samples from three (Frantoio, Leccino, Moraiolo) of the previously examined cultivars and 373 samples from other nine olive varieties (Minuta, Moraiolo, Nocellara del Belice, Nociara, Ortice, Ortolana, Ottobratica, Peranzana, Racioppella and Sinopolese). These new samples were harvested from 1996 to 2000 in six Italian regions (Calabria, Campania, Lazio, Molise, Puglia and Sicilia).Kennard-Stone algorithm was used to partition the samples into the training and test sets and the value of Fisher F-ratio was computed to identify the most discriminating indices in order to reduce the number of input variables.A first study, restricted to the original five cultivars only, showed that 12 variables are necessary in the best LDA model, which was able to correctly recognize 92.7% of the training samples and to correctly predict 90.6% of the test set. On the other hand, the first seven variables only were necessary to obtain a null prediction error over the test and validation set samples using BP-ANN.In a successive stage, ANNs have been used to extend the study to all the 14 cultivars (576 samples). In this case, the first 16 variables according to the value of Fisher F-ratio were included in the best classification model. This model was able to correctly recognize all the samples in the training set (RMS<0.00001) and to correctly predict all the samples in the test (RMS error=0.0008) and validation (RMS error=0.001) sets.     

Screening of oil samples on the basis of excitation-emission room-temperature phosphorescence data and multiway chemometric techniques. Introducing the second-order advantage in a classification study

Room-temperature phosphorescence excitation-emission matrices and multiway methods have been analyzed as potential tools for screening oil samples, based on full matrix information for polyaromatic hydrocarbons. Crude oils obtained from different sources of similar geographic origin, as well as light and heavy lubricating oils, were analyzed. The room-temperature phosphorescence matrix signals were processed by applying multilayer perceptron artificial neural networks, parallel factor analysis coupled to linear discriminant analysis, discriminant unfolded partial least-squares, and discriminant multidimensional partial least-squares (DN-PLS). The ability of the latter algorithm to classify the investigated oils into four categories is demonstrated. In addition, the combination of DN-PLS with residual bilinearization allows for a proper classification of oils containing unsuspected compounds not present in the training sample set. This second-order advantage concept is applied to a classification study for the first time. The employed approach is fast, avoids the use of laborious chromatographic analysis, and is relevant for oil characterization, identification, and determination of accidental spill sources.     

Classification of detectors for ion chromatography using principal components regression and linear discriminant analysis

Principal components regression (PCR) and linear discriminant analysis (LDA) have been applied to the classification of ion chromatographic detectors using information about the sample and other IC method conditions (19 attributes in total), a training set of 12 693 cases and a randomly-chosen test set of 1410 cases. Missing data was entered as a separate ‘unknown' code. When the value of each attribute was coded in a simple cardinal series (e.g., column=1, 2, 3, etc.), PCR correctly predicted the detector in 27% of the training set and 28% of the test set. By creating a variable (taking a value between 0 (absent) and 1 (present)) for each value of each attribute, the PCR prediction for both sets increased to 60%. LDA was more successful, predicting 69% of the detectors of each set, using a prior probability of the frequency of a given detector in the database, but this included zero hits for detectors that were poorly represented in the database. If equal prior probabilities were chosen the overall success rate dropped to 33% but now the classification of less frequently used detectors was improved. The ability of these numerically-oriented methods to classify discrete, non-numerical data, is surprisingly good and compares with induction methods, neural networks and expert systems reported previously.     

Discrimination between different samples of olive oil using variable selection techniques and modified fuzzy artmap neural networks

An electronic nose for classification of olive oil samples is presented. Principal component analysis and a modified fuzzy artmap neural network where applied to data acquired from 12 sensors. A custom designed variable selection technique was also used to boost performance. Ten different samples of olive oils were classified with 78% accuracy, and confusion occurred mostly between similar olive oils. Defective samples were separated from defect-free olive oil with 97% accuracy. These results show that careful variable selection, coupled to a modified fuzzy artmap algorithm, can significantly improve electronic nose performance.     

Artificial neural networks for the automated detection of trichloroethylene by passive Fourier transform infrared spectrometry

Artificial neural networks are applied to the automated classification of trichloroethylene (TCE) signatures from passive Fourier transform infrared remote sensing interferogram data. Through the use of three data collection methods, a combination of laboratory and field data is acquired that allows the methodology to be evaluated under a variety of infrared background conditions and in the presence of potentially interfering compounds such as sulfur hexafluoride, methyl ethyl ketone, acetone, carbon tetrachloride, and ammonia. To maximize the computational efficiency of the network optimization, experimental design techniques are employed to develop a training protocol for the network that takes into account the relationships among five variables that are related either to the network architecture or to the training process. This protocol is implemented for the case of a back-propagation neural network (BNN) and is used to develop an optimized network for the detection of TCE. The classification performance of the network is assessed by comparing both TCE detection capabilities and false detection rates to similar classification results obtained with the technique of piecewise linear discriminant analysis (PLDA). When applied to prediction data withheld from the optimization of both the BNN and PLDA algorithms, the BNN method is observed to outperform PLDA overall, with TCE detection rates in excess of 99% and false detection rates less than 0.5%.     

Classification of valence changes of trivalent rare earth ions in alkaline earth borates using artificial neural networks

The investigations of classification on the valence changes from RE³⁺ to RE²⁺ (RE≡Eu, Sm, Yb, Tm) in host compounds of alkaline earth borate were performed using artificial neural networks (ANNs). For comparison, the common methods of pattern recognition, such as SIMCA, KNN, Fisher discriminant analysis and stepwise discriminant analysis were adopted. A learning set consisting of 24 host compounds and a test set consisting of 12 host compounds were characterized by eight crystal structure parameters. These parameters were reduced from 8 to 4 by leaps and bounds algorithm. The recognition rates from 87.5 to 95.8% and prediction capabilities from 75.0 to 91.7% were obtained. The results provided by ANN method were better than that achieved by the other four methods.     

BP神经网络对陶瓷材料烧结性能预测的研究

用C＋＋语言建立了BP神经网络模型，并用TZP陶瓷材料的烧结性能数据进行训练及预测。预测结果表明，BP神经网络可以用于陶瓷材料的浇结性能的预测，并且精度较高。     

Absolute method for the assay of oleuropein in olive oils by atmospheric pressure chemical ionization tandem mass spectrometry

Oleuropein (OLP, 1), the active ingredient present (i) in food integrators extracted from olive leaves, (ii) in table olives, and (iii) in extra virgin olive oils is a nutraceutical whose health benefits have been widely documented. A new analytical method for its assay, which is based on the utilization of atmospheric pressure chemical ionization tandem mass spectrometry and on the use of a synthetic labeled analogue, the 4-trideuteriocarboxyoleuropein (2), as an internal standard, is presented. The results obtained with extra virgin olive oils from different cultivars and different Italian regions are discussed.     

Data analysis of stellar specklegrams with neural networks

An artificial neural network is applied to analysis of specklegrams of binary stars. Parameters of a binary star, the angular separation and the position angle, are estimated from the specklegrams by use of neural networks for each parameter. It is shown that a neural network is useful to analyze stellar specklegrams of binary stars.     

Identification and quantification of industrial grade glycerol adulteration in red wine with fourier transform infrared spectroscopy using chemometrics and artificial neural networks

Fourier transform infrared (FT-IR) single bounce micro-attenuated total reflectance (mATR) spectroscopy, combined with multivariate and artificial neural network (ANN) data analysis, was used to determine the adulteration of industrial grade glycerol in selected red wines. Red wine samples were artificially adulterated with industrial grade glycerol over the concentration range from 0.1 to 15% and calibration models were developed and validated. Single bounce infrared spectra of glycerol adulterated wine samples were recorded in the fingerprint mid-infrared region, 900-1500 cm(-1). Partial least squares (PLS) and PLS first derivatives were used for quantitative analysis (r2 = 0.945 to 0.998), while linear discriminant analysis (LDA) and canonical variate analysis (CVA) were used for classification and discrimination. The standard error of prediction (SEP) in the validation set was between 1.44 and 2.25%. Classification of glycerol adulterants in the different brands of red wine using CVA resulted in a classification accuracy in the range between 94 and 98%. Artificial neural network analysis based on the quick back propagation network (BPN) and the radial basis function network (RBFN) algorithms had classification success rates of 93% using BPN and 100% using RBFN. The genetic algorithm network was able to predict the concentrations of glycerol in wine up to an accuracy of r2 = 0.998.     

Sensitivity of infrared spectra to chemical functional groups

Spectral features from specific regions in infrared spectra of organic molecules can consistently be attributed to certain functional groups. Artificial neural networks were employed as a pattern recognition tool to elucidate the relationships between functional groups and spectral features. The ability of these network models to predict the presence and absence of a variety of functional groups was evaluated. The sensitivity of the artificial neural network over the entire infrared spectral region was used to generate a spectral factor representation of the major information associated with each functional group. The resulting sensitivity factors were utilized in a much simpler model for functional group prediction. Ultimately, the presence of a functional group was predicted based on the dot product of an unknown spectrum with the corresponding sensitivity factor. A probability based on Bayes' theorem was assigned to each of the predictions. The prediction accuracies were greater than 90% for all 13 functional groups considered in the investigation.     

基于随机森林利用差分拉曼光谱对塑料食品包装瓶的分类研究

目的 建立一种快速无损的检验塑料食品包装瓶的分析方法,提供一种快速分类模型。方法 利用差分拉曼光谱对100个塑料食品包装瓶样品进行检验,根据样品的差分拉曼特征峰可以对样品进行分类,样品可被分成聚对苯二甲酸乙二醇酯和聚丙烯两大类,对其中数目较多的第I类继续根据样品中所含填料的不同进行分类。利用贝叶斯判别、多层感知器和随机森林算法分别构建分类模型对继续分类结果进行分析验证。结果 第I类样本可继续被分为4类,贝叶斯判别结合留一交叉验证法分类正确率为71.7%,多层感知器神经网络分类模型的训练集和测试集分类正确率分别为100%和86.2%,随机森林分类模型的训练集和测试集分类正确率分别为100%和96.5%。通过比较发现,差分拉曼光谱与随机森林算法相结合可以对塑料食品包装瓶实现有效的分类。结论 该方法简单快速,样品用量少且无损样品,可为塑料食品包装品的物证鉴定提供科学依据。     

Application of artificial intelligence to modelling asphalt–rubber viscosity

The viscosity of binder is of great importance during the handling, mixing, application and compaction of asphalt in highway surfacing. This paper presents experimental data and the application of artificial intelligence techniques (statistics, artificial neural networks (ANNs) and fuzzy logic) to modelling of apparent viscosity in asphalt–rubber binders. The binders were prepared in the laboratory by varying the rubber content (RC), rubber particle size, duration and temperature of mixture in conformity with a statistical design plan. Multi-factorial analysis of variance showed that the RC has a major influence on the viscosity observed for the considered interval of parameters variation. When only limited experimental data of design matrix are available for modelling, the fuzzy logic model is the best model to be used. In addition, the combined use of ANN and multiple regression analysis improved the characteristics of the neural network.     

A hybrid AI optimization method applied to industrial processes

A hybrid artificial intelligent optimization (HAIO) method, which combines traditional classification pattern recognition (CPR) with artificial neural networks (ANN), applied to industrial processes is presented. This new method includes the fuzzy membership function of sample class, the center cluster of class, a hopeful region, inverse mapping, independent neural network modeling of classified analogy, CPR-ANN, GA (genetic algorithm)-ANN and network-based expert system etc. Some applications of the HAIO are shown.     

Ground Nephogram Recognition Algorithm Based on Selective Neural Network Ensemble

In view of the low accuracy of traditional ground nephogram recognition model, the authors put forward a k-means algorithm-acquired neural network ensemble method, which takes BP neural network ensemble model as the basis, uses k-means algorithm to choose the individual neural networks with partial diversities for integration, and builds the cloud form classification model. Through simulation experiments on ground nephogram samples, the results show that the algorithm proposed in the article can effectively improve the Classification accuracy of ground nephogram recognition in comparison with applying single BP neural network and traditional BP AdaBoost ensemble algorithm on classification of ground nephogram.