Alternative k-nearest neighbour rules in supervised pattern recognition : Part 3. Condensed nearest neighbour rules

Two nearest neighbour rules in which the training set is reduced to a limited number of representative objects (condensed nearest neighbour method) are discussed. The emphasis is on the probabilistic approach. The performance criteria of probabilistic methods are discussed and the performance of the condensed nearest neighbour method is compared with that of other probabilistic pattern recognition methods. This is done on the basis of a data set concerning the differentiation of thyroid functional states.     

Characterization of Galician (N.W. Spain) quality brand potatoes: a comparison study of several pattern recognition techniques

Authenticity is an important food quality criterion and rapid methods to guarantee it are widely demanded by food producers, processors, consumers and regulatory bodies. The objective of this work was to develop a classification system in order to confirm the authenticity of Galician potatoes with a Certified Brand of Origin and Quality (CBOQ) 'Denominación Específica: Patata de Galicia' and to differentiate them from other potatoes that did not have this CBOQ. Ten selected metals were determined by atomic spectroscopy in 102 potato samples which were divided into two categories: CBOQ and non-CBOQ potatoes. Multivariate chemometric techniques, such as cluster analysis and principal component analysis, were applied to perform a preliminary study of the data structure. Four supervised pattern recognition procedures [including linear discriminant analysis (LDA), K-nearest neighbours (KNN), soft independent modelling of class analogy (SIMCA) and multilayer feed-forward neural networks (MLF-ANN)] were used to classify samples into the two categories considered on the basis of the chemical data. Results for LDA, KNN and MLF-ANN are acceptable for the non-CBOQ class, whereas SIMCA showed better recognition and prediction abilities for the CBOQ class. A more sophisticated neural network approach performed by the combination of the self-organizing with adaptive neighbourhood network (SOAN) and MLF network was employed to optimize the classification. Using this combined method, excellent performance in terms of classification and prediction abilities was obtained for the two categories with a success rate ranging from 98 to 100%. The metal profiles provided sufficient information to enable classification rules to be developed for identifying potatoes according to their origin brand based on SOAN-MLF neural networks.     

Alternative k-nearest neighbour rules in supervised pattern recognition : Part 2. Probabilistic classification on the basis of the kNN method modified for direct density estimation

An implementation of the kNN classification method of Loftsgaarden and Quesenberry is discussed. It is a theoretically valid alternative for probabilistic pattern recognition by means of the Bayes equation. The method is illustrated on the basis of a data set containing laboratory tests for three categories of functional states of the thyroid gland and the performance is compared with those of other probabilistic pattern recognition techniques. Although the results are acceptable, the kNN method did not perform as well as some other probabilistic techniques.     

Extraction of information from large data sets by pattern recognition

The use of computers has made data collection much easier and analytical chemists increasingly wonder how to make use of all the data obtained. Pattern recognition permits to extract information present in large data sets in an automatic way.Many scientists acknowledge this fact but are rebutted by the task of learning to use pattern recognition methods. Indeed, there are many methods available and for the newcomer it is extremely difficult to make a selection. For this reason, the lecture will start by explaining the models used in pattern recognition. This will be followed by a critical discussion of advantages and disadvantages of the methods and a selection of preferred methods.

---

Gewinnung von Information aus großen Datenmengen mit Hilfe der Strukturerkennung

     

Orthogonal transformations for feature extraction in chemical pattern recognition

A crucial point in pattern recognition methods is the extraction of features to determine the pattern vectors. Orthogonal transformations, e.g., Fourier, Walsh and Haar, are investigated as preprocessing methods for feature extraction. The theoretical considerations and conclusions are compared with computational results obtained by applying different pattern recognition methods to two different but similar collections of low-resolution mass spectra.     

核酸适配体生物传感器在三聚氰胺检测中的应用进展

三聚氰胺常被非法添加到食品中,以提高食品中蛋白质的含量。但是,三聚氰胺一旦进入体内,会对人们的健康造成伤害。因此,对三聚氰胺的检测十分必要。为了弥补传统仪器检测法和免疫检测法的不足,基于核酸适配体开发了一系列新的生物传感器,用于三聚氰胺的检测。按照与三聚氰胺的不同识别机制,把这些新的生物传感器分成了四类,分别为基于多聚胸腺嘧啶DNA链和三聚氰胺识别的生物传感器、基于无嘌呤位点的三链DNA结构和三聚氰胺识别的生物传感器、基于核酸适配体和三聚氰胺识别的生物传感器、基于三聚氰胺和汞离子/铜离子等配位识别的生物传感器。本文按照上述四类方法逐个展开,对核酸适配体生物传感器在三聚氰胺检测中的应用进行了综述,并对它们的优缺点进行阐述。     

Evaluation of the super-modified simplex for use in chemical pattern recognition

A simplex optimization technique, the super-modified simplex (SMS), is evaluated for use in the pattern recognition analysis of low-resolution mass spectra. For the recognition of eleven functional group categories, the performances of SMS-derived weight vectors are shown to be comparable to those obtained by a previously developed modified simplex method. Data are presented which indicate that the SMS procedure requires fewer simplices and decreased computational time to converge to an optimized solution for the structural analysis problems investigated.     

Interpretation of analytical chemical information by pattern recognition methods-a survey

Since the late sixties, pattern recognition techniques have been used by analytical chemists to facilitate the interpretation of multivariate analytical information. Most research within the field has focused on adapting pattern recognition methods to chemical data. This has been necessary since chemical data are often complicated by the fact that distributions are unknown. Through the first decade of chemical pattern recognition, promising results have been obtained even though the data sets studied have frequently been rather small for statistical analysis. The past few years have shown that an increasing number of analytical chemists are interested in the sheer utility of pattern recognition. This can be taken as a valid sign of a useful approach. The present communication surveys this development. Those methods which have proved most useful for analytical chemical data are described in some detail, and applications within the various fields of analytical chemistry are reviewed.     

Pattern recognition in capillary electrophoresis data using dynamic programming in the wavelet domain

A novel approach for CE data analysis based on pattern recognition techniques in the wavelet domain is presented. Low-resolution, denoised electropherograms are obtained by applying several preprocessing algorithms including denoising, baseline correction, and detection of the region of interest in the wavelet domain. The resultant signals are mapped into character sequences using first derivative information and multilevel peak height quantization. Next, a local alignment algorithm is applied on the coded sequences for peak pattern recognition. We also propose 2-D and 3-D representations of the found patterns for fast visual evaluation of the variability of chemical substances concentration in the analyzed samples. The proposed approach is tested on the analysis of intracerebral microdialysate data obtained by CE and LIF detection, achieving a correct detection rate of about 85% with a processing time of less than 0.3 s per 25,000-point electropherogram. Using a local alignment algorithm on low-resolution denoised electropherograms might have a great impact on high-throughput CE since the proposed methodology will substitute automatic fast pattern recognition analysis for slow, human based time-consuming visual pattern recognition methods.     

Anwendung einiger Methoden der automatischen Zeichenerkennung auf die interpretation von Massenspektren

Some pattern recognition methods are briefly discussed from the geometric point of view: classification by distance measurements to prototypes, classification by decision-planes and classification by the nearest-neighbor-method. The last two methods give good results with regard to the automatic determination of molecular structures from low resolution mass spectra. It is emphasized, that pattern recognition methods may be useful for interpreting various types of data pools in physical and analytical chemistry.     

Discrimination of Chinese traditional soy sauces based on their physico-chemical properties

This work aimed to classify the categories (produced by different processes) and brands (obtained from different geographical origins) of Chinese soy sauces. Nine variables of physico-chemical properties (density, pH, dry matter, ashes, electric conductivity, amino nitrogen, salt, viscosity and total acidity) of 53 soy sauce samples were measured. The measured data was submitted to such pattern recognition as cluster analysis (CA), principal component analysis (PCA), discrimination partial least squares (DPLS), linear discrimination analysis (LDA) and K-nearest neighbor (KNN) to evaluate the data patterns and the possibility of differentiating Chinese soy sauces between different categories and brands. Two clusters corresponding to the two categories were obtained, and each cluster was divided into three subsets corresponding to three brands by the CA method. The variables for LDA and KNN were selected by the Fisher F-ratio approach. The prediction ability of all classifiers was evaluated by cross-validation. For the three supervised discrimination analyses, LDA and KNN gave 100% predications according to the sample category and brand.     

Classification of chromatographic data using multidimensional polynomials

Summary A pattern recognition methodology has been developed for analysis of chromatographic data. The method uses a new class of multidimensional orthogonal polynomials developed by Cohen in conjunction with a supervised learning technique. The method is applicable to any chromatographic data for which classification into two or more categories is desired. The algorithm analyzes both elution times and peak areas. An application is shown for the analysis of organic acids in ascitic fluid obtained from patients with liver disorders. Classification of these patients for presence or absence of bacterial infection shows over ninety percent correct classification.     

Potential methods in pattern recognition : Part 1. Classification aspects of the supervised method ALLOC

This paper on the application of potential functions in pattern recognition introduces the software package ALLOC to analytical chemistry, emphasizing the methodology of classifying objects. ALLOC is compared with other classification techniques on the basis of two data sets and is shown to perform very well.     

Extraction of information from large data sets by pattern recognition

Summary Pattern recognition permits to extract information present in large data sets in an automatic way. Many scientists acknowledge this fact but are rebutted by the task of learning to use pattern recognition methods. Indeed, there are many methods available and for the newcomer it is extremely difficult to make a selection. For this reason, the paper starts by explaining the models used in pattern recognition. This is followed by a critical discussion of advantages and disadvantages of the methods and a selection of preferred methods.

---

Gewinnung von Information aus großen Datenmengen mit Hilfe der Mustererkennung

Zusammenfassung Mit Hilfe der Mustererkennung können Informationen aus großen Datenmengen automatisch gewonnen werden. Obwohl man sich dessen allgemein bewußt ist, schreckt man doch gewöhnlich vor der Aufgabe zurück, sich mit den entsprechenden Methoden befassen zu müssen, denn es gibt sehr viele davon und eine geeignete Auswahl ist schwer zu treffen. Aus diesem Grund werden in dieser Arbeit die einzelnen Verfahren der Mustererkennung erklärt, deren Vor- und Nachteile diskutiert und eine entsprechende Auswahl geboten.

     

Use of computerized pattern recognition in the study of the cuticular hydrocarbons of imported fire ants. I. Introduction and characterization of the cuticular hydrocarbon patterns of Solenopsis invicta and S. richteri

A method is described in which gas chromatographic (GC) data obtained from cuticular hydrocarbons are treated by methods of pattern recognition. Based on a recently described sample preparation procedure, GC data are normalized to eliminate slight variations in chromatographic conditions and converted into the proper format for discriminant analysis by computer. The results of several methods of data treatment and display are discussed, based upon the chemometric system package, ARTHUR. The approach has the advantage of largely removing operator bias.     

Fingerprinting of complex mixtures with the use of high performance liquid chromatography, inductively coupled plasma atomic emission spectroscopy and chemometrics

The molecular and metal profile fingerprints were obtained from a complex substance, Atractylis chinensis DC—a traditional Chinese medicine (TCM), with the use of the high performance liquid chromatography (HPLC) and inductively coupled plasma atomic emission spectroscopy (ICP-AES) techniques. This substance was used in this work as an example of a complex biological material, which has found application as a TCM. Such TCM samples are traditionally processed by the Bran, Cut, Fried and Swill methods, and were collected from five provinces in China. The data matrices obtained from the two types of analysis produced two principal component biplots, which showed that the HPLC fingerprint data were discriminated on the basis of the methods for processing the raw TCM, while the metal analysis grouped according to the geographical origin. When the two data matrices were combined into a one two-way matrix, the resulting biplot showed a clear separation on the basis of the HPLC fingerprints. Importantly, within each different grouping the objects separated according to their geographical origin, and they ranked approximately in the same order in each group. This result suggested that by using such an approach, it is possible to derive improved characterisation of the complex TCM materials on the basis of the two kinds of analytical data.

In addition, two supervised pattern recognition methods, K-nearest neighbors (KNNs) method, and linear discriminant analysis (LDA), were successfully applied to the individual data matrices—thus, supporting the PCA approach.     

The Use of UV-Visible Diffuse Reflectance Spectrophotometry for a Fast,Preliminary Authentication of Gemstones

The identification of gemstones is an important topic in the field of cultural heritage, given their enormous value. Particularly, the most important precious stones, namely diamond, emerald, ruby and sapphire, are frequently subjected to counterfeit by substitution with objects of lesser value with similar appearance, colour or shape. While a gemmologist is able to recognise a counterfeit in most instances, more generally, it is not easy to do this without resorting to instrumental methods. In this work, the use of UV-visible diffuse reflectance spectrophotometry with optic fibres (FORS) is proposed as a fast and easy method for the preliminary identification of gemstones, alternative to the classical methods used by gemmologists or to Raman spectroscopy, which is by far the instrumental method with the best diagnostic potential, but it cannot be used in situations of problematic geometric hindrance. The possibilities and the limitations given by the FORS technique are critically discussed together with the spectral features of the most important gemstones. Finally, the application of chemometric pattern recognition methods is described for the treatment of large sets of spectral data deriving from gemstones identification.     

Effect of scaling on class modeling with the simca method

The first step in multivariate analysis is almost always the scaling of the variables. The pattern recognition technique SIMCA provides the possibility of scaling the variables over all the objects of the training set (classical scaling), or only over the objects belonging to the same group (separate scaling). The former method of scaling is the more used. The effect of separate scaling on the classification of objects with SIMCA is investigated for a data set consisting of the percentage distribution of fatty acids in olive oils originating from two neighbouring regions in Italy. It is shown that separate scaling has a beneficial effect on the classification.