Classification of analytical images with radial basis function networks and forward selection

Stubbings, T.; Hutter, Herbert

doi:10.1016/s0169-7439(99)00035-0

Cited by 22 publications

(5 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The topology and learning parameters of RBFNs are easy to optimize [4,5]. Although recently, RBFNs have been successfully applied in many multivariate calibration [6], classification [7] and QSPR studies [8,9], the present study is among the limited number of its applications in QSAR studies.…”

Section: Introductionmentioning

confidence: 98%

Correlation weighted successive projections algorithm as a novel method for variable selection in QSAR studies: investigation of anti‐HIV activity of HEPT derivatives

Kompany‐Zareh

Akhlaghi

2007

Journal of Chemometrics

View full text Add to dashboard Cite

Correlation weighted successive projections algorithm (CWSPA), as a modified version of successive projections algorithm (SPA), is proposed for selection of descriptors in the non-linear quantitative structure-activity relationship (QSAR) study of a series of 1-[2-hydroxyethoxy-methyl]-6-(phenylthio)thymine] (HEPT) derivatives, as non-nucloside reverse transcriptase inhibitors (NNRTIs). In the proposed procedure the correlation coefficient of each descriptor with the activities (r g ) was an additional criterion for selection of descriptors. The extent of contribution of r in the selection of variables, m, was also optimized and r 4 g -CWSPA was the selected condition (m ¼ 4). Three layer radial basis function networks (RBFNs) and molecular descriptors derived solely from molecular structure were used to construct the non-linear QSAR models. Utilizing r 4 g -CWSPA a limited number of uncorrelated and informative descriptors were selected. The relative standard error percent in anti-HIV activity predictions for the training set by the application of cross-validation (RSECV%) was 9.77%, and for prediction set (RSEP%) was 8.61% when the selected number of descriptors were 20. The obtained model outperforms those given in the literature in both the fitting and predicting stages. RBFN analysis yielded predicted activities in an acceptable agreement with the experimentally obtained values (cross-validation r ¼ 0.924, prediction r ¼ 0.939). Compared to SPA, r m g -CWSPA resulted in a lower RSECV% and RSEP% values using lower number of selected variables. The results show that considering the correlation of variables to the independent variables increase the performance of selection, as a result, the quality of the set of selected variables. Finally, a simple procedure for selection of variables using r 4 g -CWSPA was proposed in which there was no need to test all possible initial descriptors. The results from the simple procedure were comparable to the procedure in which all of the possible initial descriptors were tested. The proposed method was successfully validated by five different training and test sets.

show abstract

Section: Introductionmentioning

confidence: 98%

Correlation weighted successive projections algorithm as a novel method for variable selection in QSAR studies: investigation of anti‐HIV activity of HEPT derivatives

Kompany‐Zareh

Akhlaghi

2007

Journal of Chemometrics

View full text Add to dashboard Cite

show abstract

“…Characteristic absorbency values of infrared spectrum were allocated to hidden layer by input layer. The nerve ceil basis function (gauss function) of hidden layer carry out nonlinear operation on input variable, nerve ceils of output layer linear weighted array the output of hidden layer, produce one classification information which can be used to identify the quality of extraction liquid [7,8 ] .…”

Section: The Application Of Infra-red Spectrum Analysis In This Systemmentioning

confidence: 99%

The remote supervisory and controlling experiment system of traditional Chinese medicine production based on Fieldbus

Zhan

2006

Sixth International Symposium on Instrumentation and Control Technology: Sensors, Automatic Measurement, Control, and Computer

View full text Add to dashboard Cite

Since the quality of traditional Chinese medicine products are affected by raw material, machining and many other factors, it is difficult for traditional Chinese medicine production process especially the extracting process to ensure the steady and homogeneous quality. At the same time, there exist some quality control blind spots due to lacking on-line quality detection means. But if infrared spectrum analysis technology was used in traditional Chinese medicine production process on the basis of off-line analysis to real-time detect the quality of semi-manufactured goods and to be assisted by advanced automatic control technique, the steady and homogeneous quality can be obtained. It can be seen that the on-line detection of extracting process plays an important role in the development of Chinese patent medicines industry. In this paper, the design and implement of a traditional Chinese medicine extracting process monitoring experiment system which is based on PROFIBUS-DP field bus, OPC, and Internet technology is introduced. The system integrates intelligence node which gathering data, superior sub-system which achieving figure configuration and remote supervisory, during the process of traditional Chinese medicine production, monitors the temperature parameter, pressure parameter, quality parameter etc. And it can be controlled by the remote nodes in the VPN (Visual Private Network). Experiment and application do have proved that the system can reach the anticipation effect fully, and with the merits of operational stability, real-time, reliable, convenient and simple manipulation and so on.

show abstract

“…Encontrar o melhor conjunto, por "tentativa e erro" é praticamente inviável, visto que existem 2 n -1 sub-conjuntos distintos em um conjunto com n RBFs, de modo que geralmente utiliza-se de métodos heurísticos para realizar essa tarefa. Um dos tipos mais simples dentre os métodos heurísticos de otimização é a "Forward Selection" 25,27 . O modelo de otimização por "forward selection" baseia-se em iniciar um sub-conjunto vazio de RBFs e ir adicionando RBFs de modo a diminuir (minimizar) o somatório do erro quadrático ou outro critério de minimização como a validação cruzada generalizada (por exemplo).…”

Section: Figura 11 Perfil Dos Tipos De Rbf Mais Utilizados: Gaussianunclassified

Redes neurais e suas aplicações em calibração multivariada

Cerqueira¹,

Andrade²,

Poppi³

et al. 2001

Quím. Nova

View full text Add to dashboard Cite

Recebido em 28/8/00; aceito em 13/3/01 NEURAL NETWORKS AND ITS APPLICATIONS IN MULTIVARIATE CALIBRATION. Neural Networks are a set of mathematical methods and computer programs designed to simulate the information process and the knowledge acquisition of the human brain. In last years its application in chemistry is increasing significantly, due the special characteristics for model complex systems. The basic principles of two types of neural networks, the multi-layer perceptrons and radial basis functions, are introduced, as well as, a pruning approach to architecture optimization. Two analytical applications based on near infrared spectroscopy are presented, the first one for determination of nitrogen content in wheat leaves using multi-layer perceptrons networks and second one for determination of BRIX in sugar cane juices using radial basis functions networks.Keywords: neural networks; backpropagation; radial basis functions; multivariate calibration. Quim. Nova, Vol. 24, No. 6, 864-873, 2001. Divulgação INTRODUÇÃOAtualmente, tem-se observado um grande aumento na disponibilidade de informação em nosso cotidiano. A informação circula entre nós a uma velocidade que seria inconcebível até alguns anos atrás e tende aumentar ainda mais. Hoje em dia o volume de informação dobra a cada quatro anos e num futuro próximo passará a dobrar a cada dois anos. Assim, realmente vivemos no que se pode chamar de sociedade da informação. Grande parte deste aumento no volume de informações se deve ao aparecimento e uso disseminados de microcomputadores equipados com softwares, como Windows TM , que tornam o seu uso extremamente simples, até mesmo para crianças em fase de alfabetização.Uma das atividades humanas que mais se beneficiou com o uso de computadores foi a própria ciência, em particular a química analítica. Até algumas décadas atrás o principal problema enfrentado no desenvolvimento de um método analítico era como se obter os dados. Gastava-se muito tempo, dinheiro e esforço para se obter um pequeno conjunto de dados.Graças a computadorização dos instrumentos o principal problema não é mais como obter os dados, mas como tratá-los de modo eficiente. Tipicamente tais instrumentos fornecem um grande conjunto de dados e para que tais dados sejam convertidos em informação útil deve-se usar métodos matemáticos e estatísticos modernos e eficientes. Dentre tais métodos destaca-se os métodos lineares e não lineares de calibração multivariada e mais recentemente as redes neurais 1 . Para uma melhor compreensão da álgebra envolvida nos modelos de tratamento de dados, foi utilizada a seguinte notação:n Vetores -letras minúsculas em negrito (ex: x); n Matrizes -letras maiúsculas em negrito (ex: X); n Escalares -letras minúsculas sem negrito (ex: l); n Matrizes diagonais -letras maiúsculas sem negrito (ex: S). Redes NeuraisRecentes avanços em neurofisiologia têm desvendado vários mecanismos sobre o fluxo e o processamento de informações que ocorrem no cérebro humano. Alguns destes mecanismos foram modelados matematicamente permitind...

show abstract

Classification of analytical images with radial basis function networks and forward selection

Cited by 22 publications

References 8 publications

Correlation weighted successive projections algorithm as a novel method for variable selection in QSAR studies: investigation of anti‐HIV activity of HEPT derivatives

Correlation weighted successive projections algorithm as a novel method for variable selection in QSAR studies: investigation of anti‐HIV activity of HEPT derivatives

The remote supervisory and controlling experiment system of traditional Chinese medicine production based on Fieldbus

Redes neurais e suas aplicações em calibração multivariada

Contact Info

Product

Resources

About