Interpretación geométrica de redes neuronales recurrentes discretas mediante grafos completos

El objetivo del presente trabajo de investigación es explorar nuevas técnicas de implementación, basadas en grafos, para las Redes de Neuronas, con el fin de simplificar y optimizar las arquitecturas y la complejidad computacional de las mismas. Hemos centrado nuestra atención en una clase de Red de Neuronas: las Redes de Neuronas Recursivas (RNR), también conocidas como redes de Hopfield. El problema de obtener la matriz sináptica asociada con una RNR imponiendo un determinado número de vectores como puntos fijos, no está en absoluto resuelto, el número de vectores prototipo que pueden ser almacenados en la red, cuando se utiliza la ley de Hebb, es bastante limitado, la red se satura rápidamente cuando se pretende almacenar nuevos prototipos. La ley de Hebb necesita, por tanto, ser revisada. Algunas aproximaciones dirigidas a solventar dicho problema, han sido ya desarrolladas. Nosotros hemos desarrollado una nueva aproximación en la forma de implementar una RNR en orden a solucionar estos problemas. La matriz sináptica es obtenida mediante la superposición de las componentes de los vectores prototipo, sobre los vértices de un Grafo, lo cual puede ser también interpretado como una coloración de dicho grafo. Cuando el periodo de entrenamiento se termina, la matriz de adyacencia del Grafo Resultante o matriz de pesos, presenta ciertas propiedades por las cuales dichas matrices serán llamadas tetraédricas. La energía asociada a cualquier estado de la red es representado por un punto (a,b) de R2. Cada uno de los puntos de energía asociados a estados que disten lo mismo del vector cero está localizado sobre la misma línea de energía de R2. El espacio de vectores de estado puede, por tanto, clasificarse en n clases correspondientes a cada una de las n diferentes distancias que puede tener cualquier vector al vector cero. La matriz (n x n) de pesos puede reducirse a un n-vector; de esta forma, tanto el tiempo de computación como el espacio de memoria requerido par almacenar los pesos, son simplificados y optimizados. En la etapa de recuperación, es introducido un vector de parámetros R2, éste es utilizado para controlar la capacidad de la red: probaremos que lo mayor es la componente a¡, lo menor es el número de puntos fijos pertenecientes a la línea de energía R¡. Una vez que la capacidad de la red ha sido controlada mediante este parámetro, introducimos otro parámetro, definido como la desviación del vector de pesos relativos, este parámetro sirve para disminuir ostensiblemente el número de parásitos. A lo largo de todo el trabajo, hemos ido desarrollando un ejemplo, el cual nos ha servido para ir corroborando los resultados teóricos, los algoritmos están escritos en un pseudocódigo, aunque a su vez han sido implamentados utilizando el paquete Mathematica 2.2., mostrándolos en un volumen suplementario al texto.---ABSTRACT---The aim of the present research is intended to explore new specifícation techniques of Neural Networks based on Graphs to be used in the optimization and simplification of Network Architectures and Computational Complexhy. We have focused our attention in a, well known, class of Neural Networks: the Recursive Neural Networks, also known as Hopfield's Neural Networks. The general problem of constructing the synaptic matrix associated with a Recursive Neural Network imposing some vectors as fixed points is fer for completery solved, the number of prototype vectors (learning patterns) which can be stored by Hebb's law is rather limited and the memory will thus quickly reach saturation if new prototypes are continuously acquired in the course of time. Hebb's law needs thus to be revised in order to allow new prototypes to be stored at the expense of the older ones. Some approaches related with this problem has been developed. We have developed a new approach of implementing a Recursive Neural Network in order to sob/e these kind of problems, the synaptic matrix is obtained superposing the components of the prototype vectors over the vértices of a Graph which may be interpreted as a coloring of the Graph. When training is finished the adjacency matrix of the Resulting Graph or matrix of weights presents certain properties for which it may be called a tetrahedral matrix The energy associated to any possible state of the net is represented as a point (a,b) in R2. Every one of the energy points associated with state-vectors having the same Hamming distance to the zero vector are located over the same energy Une in R2. The state-vector space may be then classified in n classes according to the n different possible distances firom any of the state-vectors to the zero vector The (n x n) matrix of weights may also be reduced to a n-vector of weights, in this way the computational time and the memory space required for obtaining the weights is optimized and simplified. In the recall stage, a parameter vectora is introduced, this parameter is used for controlling the capacity of the net: it may be proved that the bigger is the r, component of J, the lower is the number of fixed points located in the r¡ energy line. Once the capacity of the net has been controlled by the ex parameter, we introduced other parameter, obtained as the relative weight vector deviation parameter, in order to reduce the number of spurious states. All along the present text, we have also developed an example, which serves as a prove for the theoretical results, the algorithms are shown in a pseudocode language in the text, these algorithm so as the graphics have been developed also using the Mathematica 2.2. mathematical package which are shown in a supplementary volume of the text.

A class of self-stabilizing MCA learning algorithms

In this letter, we propose a class of self-stabilizing learning algorithms for minor component analysis (MCA), which includes a few well-known MCA learning algorithms. Self-stabilizing means that the sign of the weight vector length change is independent of the presented input vector. For these algorithms, rigorous global convergence proof is given and the convergence rate is also discussed. By combining the positive properties of these algorithms, a new learning algorithm is proposed which can improve the performance. Simulations are employed to confirm our theoretical results.

Threshold-induced phase transitions in perceptrons

Error rates of a Boolean perceptron with threshold and either spherical or Ising constraint on the weight vector are calculated for storing patterns from biased input and output distributions derived within a one-step replica symmetry breaking (RSB) treatment. For unbiased output distribution and non-zero stability of the patterns, we find a critical load, α p, above which two solutions to the saddlepoint equations appear; one with higher free energy and zero threshold and a dominant solution with non-zero threshold. We examine this second-order phase transition and the dependence of α p on the required pattern stability, κ, for both one-step RSB and replica symmetry (RS) in the spherical case and for one-step RSB in the Ising case.

Applying SVMs and weight-based factor analysis to unsupervised adaptation for speaker verification

This paper presents an extended study on the implementation of support vector machine(SVM) based speaker verification in systems that employ continuous progressive model adaptation using the weight-based factor analysis model. The weight-based factor analysis model compensates for session variations in unsupervised scenarios by incorporating trial confidence measures in the general statistics used in the inter-session variability modelling process. Employing weight-based factor analysis in Gaussian mixture models (GMM) was recently found to provide significant performance gains to unsupervised classification. Further improvements in performance were found through the integration of SVM-based classification in the system by means of GMM supervectors. This study focuses particularly on the way in which a client is represented in the SVM kernel space using single and multiple target supervectors. Experimental results indicate that training client SVMs using a single target supervector maximises performance while exhibiting a certain robustness to the inclusion of impostor training data in the model. Furthermore, the inclusion of low-scoring target trials in the adaptation process is investigated where they were found to significantly aid performance.

Predicting residue-wise contact orders in proteins by support vector regression

Background The residue-wise contact order (RWCO) describes the sequence separations between the residues of interest and its contacting residues in a protein sequence. It is a new kind of one-dimensional protein structure that represents the extent of long-range contacts and is considered as a generalization of contact order. Together with secondary structure, accessible surface area, the B factor, and contact number, RWCO provides comprehensive and indispensable important information to reconstructing the protein three-dimensional structure from a set of one-dimensional structural properties. Accurately predicting RWCO values could have many important applications in protein three-dimensional structure prediction and protein folding rate prediction, and give deep insights into protein sequence-structure relationships. Results We developed a novel approach to predict residue-wise contact order values in proteins based on support vector regression (SVR), starting from primary amino acid sequences. We explored seven different sequence encoding schemes to examine their effects on the prediction performance, including local sequence in the form of PSI-BLAST profiles, local sequence plus amino acid composition, local sequence plus molecular weight, local sequence plus secondary structure predicted by PSIPRED, local sequence plus molecular weight and amino acid composition, local sequence plus molecular weight and predicted secondary structure, and local sequence plus molecular weight, amino acid composition and predicted secondary structure. When using local sequences with multiple sequence alignments in the form of PSI-BLAST profiles, we could predict the RWCO distribution with a Pearson correlation coefficient (CC) between the predicted and observed RWCO values of 0.55, and root mean square error (RMSE) of 0.82, based on a well-defined dataset with 680 protein sequences. Moreover, by incorporating global features such as molecular weight and amino acid composition we could further improve the prediction performance with the CC to 0.57 and an RMSE of 0.79. In addition, combining the predicted secondary structure by PSIPRED was found to significantly improve the prediction performance and could yield the best prediction accuracy with a CC of 0.60 and RMSE of 0.78, which provided at least comparable performance compared with the other existing methods. Conclusion The SVR method shows a prediction performance competitive with or at least comparable to the previously developed linear regression-based methods for predicting RWCO values. In contrast to support vector classification (SVC), SVR is very good at estimating the raw value profiles of the samples. The successful application of the SVR approach in this study reinforces the fact that support vector regression is a powerful tool in extracting the protein sequence-structure relationship and in estimating the protein structural profiles from amino acid sequences.

OCR Prediction Using Support Vector Machine Based on Piezocone Data

The determination of the overconsolidation ratio (OCR) of clay deposits is an important task in geotechnical engineering practice. This paper examines the potential of a support vector machine (SVM) for predicting the OCR of clays from piezocone penetration test data. SVM is a statistical learning theory based on a structural risk minimization principle that minimizes both error and weight terms. The five input variables used for the SVM model for prediction of OCR are the corrected cone resistance (qt), vertical total stress (sigmav), hydrostatic pore pressure (u0), pore pressure at the cone tip (u1), and the pore pressure just above the cone base (u2). Sensitivity analysis has been performed to investigate the relative importance of each of the input parameters. From the sensitivity analysis, it is clear that qt=primary in situ data influenced by OCR followed by sigmav, u0, u2, and u1. Comparison between SVM and some of the traditional interpretation methods is also presented. The results of this study have shown that the SVM approach has the potential to be a practical tool for determination of OCR.

Vector evaluated particle swarm optimization (VEPSO) for multi-objective design optimization of composite structures

We present a generic method/model for multi-objective design optimization of laminated composite components, based on vector evaluated particle swarm optimization (VEPSO) algorithm. VEPSO is a novel, co-evolutionary multi-objective variant of the popular particle swarm optimization algorithm (PSO). In the current work a modified version of VEPSO algorithm for discrete variables has been developed and implemented successfully for the, multi-objective design optimization of composites. The problem is formulated with multiple objectives of minimizing weight and the total cost of the composite component to achieve a specified strength. The primary optimization variables are - the number of layers, its stacking sequence (the orientation of the layers) and thickness of each layer. The classical lamination theory is utilized to determine the stresses in the component and the design is evaluated based on three failure criteria; failure mechanism based failure criteria, Maximum stress failure criteria and the Tsai-Wu failure criteria. The optimization method is validated for a number of different loading configurations - uniaxial, biaxial and bending loads. The design optimization has been carried for both variable stacking sequences, as well fixed standard stacking schemes and a comparative study of the different design configurations evolved has been presented. (C) 2007 Elsevier Ltd. All rights reserved.

MPI-based parallel synchronous vector evaluated particle swarm optimization for multi-objective design optimization of composite structures

This paper presents a decentralized/peer-to-peer architecture-based parallel version of the vector evaluated particle swarm optimization (VEPSO) algorithm for multi-objective design optimization of laminated composite plates using message passing interface (MPI). The design optimization of laminated composite plates being a combinatorially explosive constrained non-linear optimization problem (CNOP), with many design variables and a vast solution space, warrants the use of non-parametric and heuristic optimization algorithms like PSO. Optimization requires minimizing both the weight and cost of these composite plates, simultaneously, which renders the problem multi-objective. Hence VEPSO, a multi-objective variant of the PSO algorithm, is used. Despite the use of such a heuristic, the application problem, being computationally intensive, suffers from long execution times due to sequential computation. Hence, a parallel version of the PSO algorithm for the problem has been developed to run on several nodes of an IBM P720 cluster. The proposed parallel algorithm, using MPI's collective communication directives, establishes a peer-to-peer relationship between the constituent parallel processes, deviating from the more common master-slave approach, in achieving reduction of computation time by factor of up to 10. Finally we show the effectiveness of the proposed parallel algorithm by comparing it with a serial implementation of VEPSO and a parallel implementation of the vector evaluated genetic algorithm (VEGA) for the same design problem. (c) 2012 Elsevier Ltd. All rights reserved.

Efficient Dendrimer-DNA Complexation and Gene Delivery Vector Properties of Nitrogen-Core Poly(propyl ether imine) Dendrimer in Mammalian Cells

Dendrimers as vectors for gene delivery were established, primarily by utilizing few prominent dendrimer types so far. We report herein studies of DNA complexation efficacies and gene delivery vector properties of a nitrogen-core poly(propyl ether imine) (PETIM) dendrimer, constituted with 22 tertiary amine internal branches and 24 primary amines at the periphery. The interaction of the dendrimer with pEGFPDNA was evaluated through UV-vis, circular dichroism (CD) spectral studies, ethidium bromide fluorescence emission quenching, thermal melting, and gel retardation assays, from which most changes to DNA structure during complexation was found to occur at a weight ratio of dendrimer:DNA similar to 2:1. The zeta potential measurements further confirmed this stoichiometry at electroneutrality. The structure of a DNA oligomer upon dendrimer complexation was simulated through molecular modeling and the simulation showed that the dendrimer enfolded DNA oligomer along both major and minor grooves, without causing DNA deformation, in 1:1 and 2:1 dendrimer-to-DNA complexes. Atomic force microscopy (AFM) studies on dendrimer-pEGFP DNA complex showed an increase in the average z-height as a result of dendrimers decorating the DNA, without causing a distortion of the DNA structure. Cytotoxicity studies involving five different mammalian cell lines, using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide] (MTT) assay, reveal the dendrimer toxicity profile (IC50) values of similar to 400-1000 mu g mL(-1), depending on the cell line tested. Quantitative estimation, using luciferase assay, showed that the gene transfection was at least 100 times higher when compared to poly(ethylene imine) branched polymer, having similar number of cationic sites as the dendrimer. The present study establishes the physicochemical behavior of new nitrogen-core PETIM dendrimer-DNA complexes, their lower toxicities, and efficient gene delivery vector properties.

Low Molecular Weight Chitosan (LMWC)-based Polyplexes for pDNA Delivery: From Bench to Bedside

Non-viral gene delivery vectors are emerging as a safer alternative to viral vectors. Among natural polymers, chitosan (Ch) is the most studied one, and low molecular weight Ch, specifically, presents a wide range of advantages for non-viral pDNA delivery. It is crucial to determine the best process for the formation of Low Molecular Weight Chitosan (LMWC)-pDNA complexes and to characterize their physicochemical properties to better understand their behavior once the polyplexes are administered. The transfection efficiency of Ch based polyplexes is relatively low. Therefore, it is essential to understand all the transfection process, including the cellular uptake, endosomal escape and nuclear import, together with the parameters involved in the process to improve the design and development of the non-viral vectors. The aim of this review is to describe the formation and characterization of LMWC based polyplexes, the in vitro transfection process and finally, the in vivo applications of LMWC based polyplexes for gene therapy purposes.

Potential use of a recombinant replication-defective adenovirus vector carrying the C-terminal portion of the P97 adhesin protein as a vaccine against Mycoplasma hyopneumoniae in swine.

Mycoplasma hyopneumoniae causes severe economic losses to the swine industry worldwide and the prevention of its related disease, enzootic porcine pneumonia, remains a challenge. The P97 adhesin protein of M. hyopneumoniae should be a good candidate for the development of a subunit vaccine because antibodies produced against P97 could prevent the adhesion of the pathogen to the respiratory epithelial cells in vitro. In the present study, a P97 recombinant replication-defective adenovirus (rAdP97c) subunit vaccine efficiency was evaluated in pigs. The rAdP97c vaccine was found to induce both strong P97 specific humoral and cellular immune responses. The rAdP97c vaccinated pigs developed a lower amount of macroscopic lung lesions (18.5 ± 9.6%) compared to the unvaccinated and challenged animals (45.8 ± 11.5%). rAdP97c vaccine reduced significantly the severity of inflammatory response and the amount of M. hyopneumoniae in the respiratory tract. Furthermore, the average daily weight gain was slightly improved in the rAdP97c vaccinated pigs (0.672 ± 0.068 kg/day) compared to the unvaccinated and challenged animals (0.568 ± 0.104 kg/day). A bacterin-based commercial vaccine (Suvaxyn® MH-one) was more efficient to induce a protective immune response than rAdP97c even if it did not evoke a P97 specific immune response. These results suggest that immunodominant antigens other than P97 adhesin are also important in the induction of a protective immune response and should be taken into account in the future development of M. hyopneumoniae subunit vaccines.

Predicting body and carcass characteristics of 2 broiler chicken strains using support vector regression and neural network models

As a new modeling method, support vector regression (SVR) has been regarded as the state-of-the-art technique for regression and approximation. In this study, the SVR models had been introduced and developed to predict body and carcass-related characteristics of 2 strains of broiler chicken. To evaluate the prediction ability of SVR models, we compared their performance with that of neural network (NN) models. Evaluation of the prediction accuracy of models was based on the R-2, MS error, and bias. The variables of interest as model output were BW, empty BW, carcass, breast, drumstick, thigh, and wing weight in 2 strains of Ross and Cobb chickens based on intake dietary nutrients, including ME (kcal/bird per week), CP, TSAA, and Lys, all as grams per bird per week. A data set composed of 64 measurements taken from each strain were used for this analysis, where 44 data lines were used for model training, whereas the remaining 20 lines were used to test the created models. The results of this study revealed that it is possible to satisfactorily estimate the BW and carcass parts of the broiler chickens via their dietary nutrient intake. Through statistical criteria used to evaluate the performance of the SVR and NN models, the overall results demonstrate that the discussed models can be effective for accurate prediction of the body and carcass-related characteristics investigated here. However, the SVR method achieved better accuracy and generalization than the NN method. This indicates that the new data mining technique (SVR model) can be used as an alternative modeling tool for NN models. However, further reevaluation of this algorithm in the future is suggested.

Classification of petroleum well drilling operations using Support Vector Machine (SVM)

During the petroleum well drilling operation many mechanical and hydraulic parameters are monitored by an instrumentation system installed in the rig called a mud-logging system. These sensors, distributed in the rig, monitor different operation parameters such as weight on the hook and drillstring rotation. These measurements are known as mud-logging records and allow the online following of all the drilling process with well monitoring purposes. However, in most of the cases, these data are stored without taking advantage of all their potential. On the other hand, to make use of the mud-logging data, an analysis and interpretationt is required. That is not an easy task because of the large volume of information involved. This paper presents a Support Vector Machine (SVM) used to automatically classify the drilling operation stages through the analysis of some mud-logging parameters. In order to validate the results of SVM technique, it was compared to a classification elaborated by a Petroleum Engineering expert. © 2006 IEEE.

Computations of Vector-Valued Siegel Modular Forms

We carry out some computations of vector-valued Siegel modular forms of degree two, weight (k, 2) and level one, and highlight three experimental results: (1) we identify a rational eigenform in a three-dimensional space of cusp forms; (2) we observe that non-cuspidal eigenforms of level one are not always rational; (3) we verify a number of cases of conjectures about congruences between classical modular forms and Siegel modular forms. Our approach is based on Satoh's description of the module of vector-valued Siegel modular forms of weight (k, 2) and an explicit description of the Hecke action on Fourier expansions. (C) 2013 Elsevier Inc. All rights reserved.

Local targeting of malignant gliomas by the diffusible peptidic vector 1,4,7,10-tetraazacyclododecane-1-glutaric acid-4,7,10-triacetic acid-substance p

PURPOSE: Malignant glial brain tumors consistently overexpress neurokinin type 1 receptors. In classic seed-based brachytherapy, one to several rigid (125)I seeds are inserted, mainly for the treatment of small low-grade gliomas. The complex geometry of rapidly proliferating high-grade gliomas requires a diffusible system targeting tumor-associated surface structures to saturate the tumor, including its margins. EXPERIMENTAL DESIGN: We developed a new targeting vector by conjugating the chelator 1,4,7,10-tetraazacyclododecane-1-glutaric acid-4,7,10-triacetic acid to Arg(1) of substance P, generating a radiopharmaceutical with a molecular weight of 1,806 Da and an IC(50) of 0.88 +/- 0.34 nmol/L. Cell biological studies were done with glioblastoma cell lines. neurokinin type-1 receptor (NK1R) autoradiography was done with 58 tumor biopsies. For labeling, (90)Y was mostly used. To reduce the "cross-fire effect" in critically located tumors, (177)Lut and (213)Bi were used instead. In a pilot study, we assessed feasibility, biodistribution, and early and long-term toxicity following i.t. injection of radiolabeled 1,4,7,10-tetraazacyclododecane-1-glutaric acid-4,7,10-triacetic acid substance P in 14 glioblastoma and six glioma patients of WHO grades 2 to 3. RESULTS: Autoradiography disclosed overexpression of NK1R in 55 of 58 gliomas of WHO grades 2 to 4. Internalization of the peptidic vector was found to be specific. Clinically, the radiopharmeutical was distributed according to tumor geometry. Only transient toxicity was seen as symptomatic radiogenic edema in one patient (observation period, 7-66 months). Disease stabilization and/or improved neurologic status was observed in 13 of 20 patients. Secondary resection disclosed widespread radiation necrosis with improved demarcation. CONCLUSIONS: Targeted radiotherapy using diffusible peptidic vectors represents an innovative strategy for local control of malignant gliomas, which will be further assessed as a neoadjuvant approach.