Critério de correntropia no treinamento de redes fuzzy wavelet neural networks para identificação de sistemas dinâmicos não lineares

The great interest in nonlinear system identification is mainly due to the fact that a large amount of real systems are complex and need to have their nonlinearities considered so that their models can be successfully used in applications of control, prediction, inference, among others. This work evaluates the application of Fuzzy Wavelet Neural Networks (FWNN) to identify nonlinear dynamical systems subjected to noise and outliers. Generally, these elements cause negative effects on the identification procedure, resulting in erroneous interpretations regarding the dynamical behavior of the system. The FWNN combines in a single structure the ability to deal with uncertainties of fuzzy logic, the multiresolution characteristics of wavelet theory and learning and generalization abilities of the artificial neural networks. Usually, the learning procedure of these neural networks is realized by a gradient based method, which uses the mean squared error as its cost function. This work proposes the replacement of this traditional function by an Information Theoretic Learning similarity measure, called correntropy. With the use of this similarity measure, higher order statistics can be considered during the FWNN training process. For this reason, this measure is more suitable for non-Gaussian error distributions and makes the training less sensitive to the presence of outliers. In order to evaluate this replacement, FWNN models are obtained in two identification case studies: a real nonlinear system, consisting of a multisection tank, and a simulated system based on a model of the human knee joint. The results demonstrate that the application of correntropy as the error backpropagation algorithm cost function makes the identification procedure using FWNN models more robust to outliers. However, this is only achieved if the gaussian kernel width of correntropy is properly adjusted.

Ensemble Stream Model for Data-Cleaning in Sensor Networks

Ensemble Stream Modeling and Data-cleaning are sensor information processing systems have different training and testing methods by which their goals are cross-validated. This research examines a mechanism, which seeks to extract novel patterns by generating ensembles from data. The main goal of label-less stream processing is to process the sensed events to eliminate the noises that are uncorrelated, and choose the most likely model without over fitting thus obtaining higher model confidence. Higher quality streams can be realized by combining many short streams into an ensemble which has the desired quality. The framework for the investigation is an existing data mining tool. First, to accommodate feature extraction such as a bush or natural forest-fire event we make an assumption of the burnt area (BA*), sensed ground truth as our target variable obtained from logs. Even though this is an obvious model choice the results are disappointing. The reasons for this are two: One, the histogram of fire activity is highly skewed. Two, the measured sensor parameters are highly correlated. Since using non descriptive features does not yield good results, we resort to temporal features. By doing so we carefully eliminate the averaging effects; the resulting histogram is more satisfactory and conceptual knowledge is learned from sensor streams. Second is the process of feature induction by cross-validating attributes with single or multi-target variables to minimize training error. We use F-measure score, which combines precision and accuracy to determine the false alarm rate of fire events. The multi-target data-cleaning trees use information purity of the target leaf-nodes to learn higher order features. A sensitive variance measure such as f-test is performed during each node’s split to select the best attribute. Ensemble stream model approach proved to improve when using complicated features with a simpler tree classifier. The ensemble framework for data-cleaning and the enhancements to quantify quality of fitness (30% spatial, 10% temporal, and 90% mobility reduction) of sensor led to the formation of streams for sensor-enabled applications. Which further motivates the novelty of stream quality labeling and its importance in solving vast amounts of real-time mobile streams generated today.

Higher charges and regularized quantum trace identities in su(1,1) Landau-Lifshitz model

We solve the operator ordering problem for the quantum continuous integrable su(1,1) Landau-Lifshitz model, and give a prescription to obtain the quantum trace identities, and the spectrum for the higher-order local charges. We also show that this method, based on operator regularization and renormalization, which guarantees quantum integrability, as well as the construction of self-adjoint extensions, can be used as an alternative to the discretization procedure, and unlike the latter, is based only on integrable representations. (C) 2010 American Institute of Physics. [doi:10.1063/1.3509374]

Fractional Order Junctions

Gottfried Leibniz generalized the derivation and integration, extending the operators from integer up to real, or even complex, orders. It is presently recognized that the resulting models capture long term memory effects difficult to describe by classical tools. Leon Chua generalized the set of lumped electrical elements that provide the building blocks in mathematical models. His proposal of the memristor and of higher order elements broadened the scope of variables and relationships embedded in the development of models. This paper follows the two directions and proposes a new logical step, by generalizing the concept of junction. Classical junctions interconnect system elements using simple algebraic restrictions. Nevertheless, this simplistic approach may be misleading in the presence of unexpected dynamical phenomena and requires including additional “parasitic” elements. The novel γ-junction includes, as special cases, the standard series and parallel connections and allows a new degree of freedom when building models. The proposal motivates the search for experimental and real world manifestations of the abstract conjectures.

Automatic Small Bowel Tumor Diagnosis by Using Multi-Scale Wavelet-Based Analysis in Wireless Capsule Endoscopy Images

BACKGROUND: Wireless capsule endoscopy has been introduced as an innovative, non-invasive diagnostic technique for evaluation of the gastrointestinal tract, reaching places where conventional endoscopy is unable to. However, the output of this technique is an 8 hours video, whose analysis by the expert physician is very time consuming. Thus, a computer assisted diagnosis tool to help the physicians to evaluate CE exams faster and more accurately is an important technical challenge and an excellent economical opportunity. METHOD: The set of features proposed in this paper to code textural information is based on statistical modeling of second order textural measures extracted from co-occurrence matrices. To cope with both joint and marginal non-Gaussianity of second order textural measures, higher order moments are used. These statistical moments are taken from the two-dimensional color-scale feature space, where two different scales are considered. Second and higher order moments of textural measures are computed from the co-occurrence matrices computed from images synthesized by the inverse wavelet transform of the wavelet transform containing only the selected scales for the three color channels. The dimensionality of the data is reduced by using Principal Component Analysis. RESULTS: The proposed textural features are then used as the input of a classifier based on artificial neural networks. Classification performances of 93.1% specificity and 93.9% sensitivity are achieved on real data. These promising results open the path towards a deeper study regarding the applicability of this algorithm in computer aided diagnosis systems to assist physicians in their clinical practice.

Looming signals reveal synergistic principles of multisensory integration.

Multisensory interactions are a fundamental feature of brain organization. Principles governing multisensory processing have been established by varying stimulus location, timing and efficacy independently. Determining whether and how such principles operate when stimuli vary dynamically in their perceived distance (as when looming/receding) provides an assay for synergy among the above principles and also means for linking multisensory interactions between rudimentary stimuli with higher-order signals used for communication and motor planning. Human participants indicated movement of looming or receding versus static stimuli that were visual, auditory, or multisensory combinations while 160-channel EEG was recorded. Multivariate EEG analyses and distributed source estimations were performed. Nonlinear interactions between looming signals were observed at early poststimulus latencies (∼75 ms) in analyses of voltage waveforms, global field power, and source estimations. These looming-specific interactions positively correlated with reaction time facilitation, providing direct links between neural and performance metrics of multisensory integration. Statistical analyses of source estimations identified looming-specific interactions within the right claustrum/insula extending inferiorly into the amygdala and also within the bilateral cuneus extending into the inferior and lateral occipital cortices. Multisensory effects common to all conditions, regardless of perceived distance and congruity, followed (∼115 ms) and manifested as faster transition between temporally stable brain networks (vs summed responses to unisensory conditions). We demonstrate the early-latency, synergistic interplay between existing principles of multisensory interactions. Such findings change the manner in which to model multisensory interactions at neural and behavioral/perceptual levels. We also provide neurophysiologic backing for the notion that looming signals receive preferential treatment during perception.

Emotionally negative stimuli can overcome attentional deficits in patients with visuo-spatial hemineglect.

Left unilateral spatial neglect resulting from right brain damage is characterized by loss of awareness for stimuli in the contralesional side of space, despite intact visual pathways. We examined using fMRI whether patients with neglect are more likely to consciously detect in the neglected hemifield, emotionally negative complex scenes rather than visually similar neutral pictures and if so, what neural mechanisms mediate this effect. Photographs of emotional and neutral scenes taken from the IAPS were presented in a divided visual field paradigm. As expected, the detection rate for emotional stimuli presented in the neglected field was higher than for neutral ones. Successful detection of emotional scenes as opposed to neutral stimuli in the left visual field (LVF) produced activations in the parahippocampal and anterior cingulate areas in the right hemisphere. Detection of emotional stimuli presented in the intact right visual field (RVF) activated a distributed network of structures in the left hemisphere, including anterior and posterior cingulate cortex, insula, as well as visual striate and extrastriate areas. LVF-RVF contrasts for emotional stimuli revealed activations in right and left attention related prefrontal areas whereas RVF-LVF comparison showed activations in the posterior cingulate and extrastriate visual cortex in the left hemisphere. An additional analysis contrasting detected vs. undetected emotional LVF stimuli showed involvement of left anterior cingulate, right frontal and extrastriate areas. We hypothesize that beneficial role of emotion in overcoming neglect is achieved by activation of frontal and limbic lobe networks, which provide a privileged access of emotional stimuli to attention by top-down modulation of processing in the higher-order extrastriate visual areas. Our results point to the importance of top-down regulatory role of the frontal attentional systems, which might enhance visual activations and lead to greater salience of emotional stimuli for perceptual awareness.

Towards robust network based complex systems : from evolutionary cellular automata to biological models

Abstract Sitting between your past and your future doesn't mean you are in the present. Dakota Skye Complex systems science is an interdisciplinary field grouping under the same umbrella dynamical phenomena from social, natural or mathematical sciences. The emergence of a higher order organization or behavior, transcending that expected of the linear addition of the parts, is a key factor shared by all these systems. Most complex systems can be modeled as networks that represent the interactions amongst the system's components. In addition to the actual nature of the part's interactions, the intrinsic topological structure of underlying network is believed to play a crucial role in the remarkable emergent behaviors exhibited by the systems. Moreover, the topology is also a key a factor to explain the extraordinary flexibility and resilience to perturbations when applied to transmission and diffusion phenomena. In this work, we study the effect of different network structures on the performance and on the fault tolerance of systems in two different contexts. In the first part, we study cellular automata, which are a simple paradigm for distributed computation. Cellular automata are made of basic Boolean computational units, the cells; relying on simple rules and information from- the surrounding cells to perform a global task. The limited visibility of the cells can be modeled as a network, where interactions amongst cells are governed by an underlying structure, usually a regular one. In order to increase the performance of cellular automata, we chose to change its topology. We applied computational principles inspired by Darwinian evolution, called evolutionary algorithms, to alter the system's topological structure starting from either a regular or a random one. The outcome is remarkable, as the resulting topologies find themselves sharing properties of both regular and random network, and display similitudes Watts-Strogtz's small-world network found in social systems. Moreover, the performance and tolerance to probabilistic faults of our small-world like cellular automata surpasses that of regular ones. In the second part, we use the context of biological genetic regulatory networks and, in particular, Kauffman's random Boolean networks model. In some ways, this model is close to cellular automata, although is not expected to perform any task. Instead, it simulates the time-evolution of genetic regulation within living organisms under strict conditions. The original model, though very attractive by it's simplicity, suffered from important shortcomings unveiled by the recent advances in genetics and biology. We propose to use these new discoveries to improve the original model. Firstly, we have used artificial topologies believed to be closer to that of gene regulatory networks. We have also studied actual biological organisms, and used parts of their genetic regulatory networks in our models. Secondly, we have addressed the improbable full synchronicity of the event taking place on. Boolean networks and proposed a more biologically plausible cascading scheme. Finally, we tackled the actual Boolean functions of the model, i.e. the specifics of how genes activate according to the activity of upstream genes, and presented a new update function that takes into account the actual promoting and repressing effects of one gene on another. Our improved models demonstrate the expected, biologically sound, behavior of previous GRN model, yet with superior resistance to perturbations. We believe they are one step closer to the biological reality.

Semi-Parametric Weak Instrument Regressions with an Application to the Risk-Return Trade-off

Recent work shows that a low correlation between the instruments and the included variables leads to serious inference problems. We extend the local-to-zero analysis of models with weak instruments to models with estimated instruments and regressors and with higher-order dependence between instruments and disturbances. This makes this framework applicable to linear models with expectation variables that are estimated non-parametrically. Two examples of such models are the risk-return trade-off in finance and the impact of inflation uncertainty on real economic activity. Results show that inference based on Lagrange Multiplier (LM) tests is more robust to weak instruments than Wald-based inference. Using LM confidence intervals leads us to conclude that no statistically significant risk premium is present in returns on the S&P 500 index, excess holding yields between 6-month and 3-month Treasury bills, or in yen-dollar spot returns.

Training deep convolutional architectures for vision

Les tâches de vision artificielle telles que la reconnaissance d’objets demeurent irrésolues à ce jour. Les algorithmes d’apprentissage tels que les Réseaux de Neurones Artificiels (RNA), représentent une approche prometteuse permettant d’apprendre des caractéristiques utiles pour ces tâches. Ce processus d’optimisation est néanmoins difficile. Les réseaux profonds à base de Machine de Boltzmann Restreintes (RBM) ont récemment été proposés afin de guider l’extraction de représentations intermédiaires, grâce à un algorithme d’apprentissage non-supervisé. Ce mémoire présente, par l’entremise de trois articles, des contributions à ce domaine de recherche. Le premier article traite de la RBM convolutionelle. L’usage de champs réceptifs locaux ainsi que le regroupement d’unités cachées en couches partageant les même paramètres, réduit considérablement le nombre de paramètres à apprendre et engendre des détecteurs de caractéristiques locaux et équivariant aux translations. Ceci mène à des modèles ayant une meilleure vraisemblance, comparativement aux RBMs entraînées sur des segments d’images. Le deuxième article est motivé par des découvertes récentes en neurosciences. Il analyse l’impact d’unités quadratiques sur des tâches de classification visuelles, ainsi que celui d’une nouvelle fonction d’activation. Nous observons que les RNAs à base d’unités quadratiques utilisant la fonction softsign, donnent de meilleures performances de généralisation. Le dernière article quand à lui, offre une vision critique des algorithmes populaires d’entraînement de RBMs. Nous montrons que l’algorithme de Divergence Contrastive (CD) et la CD Persistente ne sont pas robustes : tous deux nécessitent une surface d’énergie relativement plate afin que leur chaîne négative puisse mixer. La PCD à "poids rapides" contourne ce problème en perturbant légèrement le modèle, cependant, ceci génère des échantillons bruités. L’usage de chaînes tempérées dans la phase négative est une façon robuste d’adresser ces problèmes et mène à de meilleurs modèles génératifs.

Use of cellular impedance to characterize ligand functional selectivity at G protein-coupled receptors

Les récepteurs couplés aux protéines G (RCPGs) représentent la plus grande famille de cibles thérapeutiques pour le traitement d’une panoplie de pathologies humaines. Bien que plusieurs décennies de recherche aient permis de façonner nos connaissances sur ces protéines membranaires, notre compréhension des déterminants moléculaires de leur activité signalétique reste encore limitée. De ces domaines de recherche, une avancée récente a mis à jour un nouveau phénomène, appelé sélectivité fonctionnelle des ligands, qui a bouleversé les paradigmes décrivant leu fonctionnement de ces récepteurs. Ce concept émane d’observations montrant que l’activité pharmacologique de certains ligands n’est pas nécessairement conservée sur tout le répertoire signalétiques connu du récepteur et peu se restreindre à l'activation sélective d’un sous-groupe de voies de signalisation.Ce nouveau modèle pharmacologique de l'activation des RCPG ouvre de nouvelles possibilités pour la découverte de médicaments plus efficace et sûr, ciblant les RCPGs. En effet, il permet la conception de molécules modulant spécifiquement les voies signalétiques d’intérêt thérapeutique, sans engager les autres voies qui pourraient mener à des effets secondaires indésirables ou de la tolérance. Cette thèse décrit l'utilisation d'une nouvelle approche sans marquage, basée sur la mesure du changement l'impédance cellulaire. Par la mesure des changements cellulaires, comme la morphologie, l’adhésion et/ou la redistribution des macromolécules, cette approche permet de mesurer de façon simultanée l'activité de plusieurs voies de signalisation impliqués dans ces réponses. Utilisant le récepteur β2-adrénergique (β2AR) comme modèle, nous avons démontré que les variations dans l’impédance cellulaire étaient directement liées à l’activation de multiples voies de signalisation suite à la stimulation du récepteur par son ligand. L’agoniste type du β2AR, l’isoprotérénol, s’est avéré induire une réponse d’impédance dose-dépendante constituée, dans le temps, de plusieurs caractéristiques distinctes pouvant être bloquées de façon compétitive par l’antagoniste ICI118,551 Par l’utilisation d’inhibiteurs sélectifs, nous avons été en mesure de déterminer la contribution de plusieurs voies signalétiques canoniques, comme les voies dépendantes de Gs et Gi, la production d’AMPc et l’activation de ERK1/2, sur ces changements. De plus, la dissection de la réponse d’impédance a permis d’identifier une nouvelle voie de mobilisation du Ca2+ contribuant à la réponse globale des changements initiés par la stimulation du β2AR. Dans une autre étude, nous avons rapporté que la réponse calcique induite par le β2AR serait attribuable à une transactivation Gs-dépendant du récepteur purinergique P2Y11, lui-même couplé à la protéine Gq. La mesure d’impédance permettant de distinguer et de décrire une pléiade d’activités signalétiques, nous avons émis l’hypothèse que des ligands arborant des profils signalétiques différents généreraient des réponses d’impédance distinctes. Le criblage d’une librairie de ligands spécifiques au β2AR a révélé une grande variété de signatures d’impédance. Grâce au développement d’une approche computationnelle innovatrice, nous avons été en mesure de regrouper ces signatures en cinq classes de composés, un regroupement qui s’est avéré hautement corrélé avec le profil signalétique des différents ligands. Nous avons ensuite combiné le criblage de composés par impédance avec l’utilisation d’inhibiteurs sélectifs de voies signalétiques afin d’augmenter la résolution du regroupement. En évaluant l’impact d’une voie signalétique donnée sur la signature d’impédance, nous avons été en mesure de révéler une plus grande variété de textures parmi les ligands. De plus, cette méthode s’est avérée efficace pour prédire le profil signalétique d’une librairie de composés non caractérisés, ciblant le β2AR. Ces travaux ont mené à l’élaboration d’une méthode permettant d’exprimer visuellement la sélectivité fonctionnelle de ligands et ont révélé de nouvelles classes de composés pour ce récepteur. Ces nouvelles classes de composés ont ensuite été testées sur des cardiomyocytes humains, confirmant que les composés regroupés dans différentes classes produisent des effets distincts sur la contractilité de ces cellules. Globalement, ces travaux démontrent la pertinence de l’utilisation de l’impédance cellulaire pour une évaluation précise des différences fonctionnelles parmi les composés ciblant les RCPGs. En fournissant une représentation pluridimensionnelle de la signalisation émanant des RCPGs à l’aide d’un seul essai ne requérant pas de marquage, les signatures d’impédance représentent une stratégie simple et innovante pour l’évaluation de la fonctionnalité sélective des ligands. Cette méthode pourrait être d’une grande utilité dans le processus de découverte de nouveaux médicaments.

Implementation of a Neural Network Classifier for Noise Sources in the Ocean

The paper investigates the feasibility of implementing an intelligent classifier for noise sources in the ocean, with the help of artificial neural networks, using higher order spectral features. Non-linear interactions between the component frequencies of the noise data can give rise to certain phase relations called Quadratic Phase Coupling (QPC), which cannot be characterized by power spectral analysis. However, bispectral analysis, which is a higher order estimation technique, can reveal the presence of such phase couplings and provide a measure to quantify such couplings. A feed forward neural network has been trained and validated with higher order spectral features

Properties of Equilibrium Distributions of Order n

The present work is intended to discuss various properties and reliability aspects of higher order equilibrium distributions in continuous, discrete and multivariate cases, which contribute to the study on equilibrium distributions. At first, we have to study and consolidate the existing literature on equilibrium distributions. For this we need some basic concepts in reliability. These are being discussed in the 2nd chapter, In Chapter 3, some identities connecting the failure rate functions and moments of residual life of the univariate, non-negative continuous equilibrium distributions of higher order and that of the baseline distribution are derived. These identities are then used to characterize the generalized Pareto model, mixture of exponentials and gamma distribution. An approach using the characteristic functions is also discussed with illustrations. Moreover, characterizations of ageing classes using stochastic orders has been discussed. Part of the results of this chapter has been reported in Nair and Preeth (2009). Various properties of equilibrium distributions of non-negative discrete univariate random variables are discussed in Chapter 4. Then some characterizations of the geo- metric, Waring and negative hyper-geometric distributions are presented. Moreover, the ageing properties of the original distribution and nth order equilibrium distribu- tions are compared. Part of the results of this chapter have been reported in Nair, Sankaran and Preeth (2012). Chapter 5 is a continuation of Chapter 4. Here, several conditions, in terms of stochastic orders connecting the baseline and its equilibrium distributions are derived. These conditions can be used to rede_ne certain ageing notions. Then equilibrium distributions of two random variables are compared in terms of various stochastic orders that have implications in reliability applications. In Chapter 6, we make two approaches to de_ne multivariate equilibrium distribu- tions of order n. Then various properties including characterizations of higher order equilibrium distributions are presented. Part of the results of this chapter have been reported in Nair and Preeth (2008). The Thesis is concluded in Chapter 7. A discussion on further studies on equilib- rium distributions is also made in this chapter.

Low-Cost Micromechanically Tunable Optical Devices: Strained Resonator Engineering, Technological Implementation and Characterization

The rapid growth of the optical communication branches and the enormous demand for more bandwidth require novel networks such as dense wavelength division multiplexing (DWDM). These networks enable higher bitrate transmission using the existing optical fibers. Micromechanically tunable optical microcavity devices like VCSELs, Fabry-Pérot filters and photodetectors are core components of these novel DWDM systems. Several air-gap based tunable devices were successfully implemented in the last years. Even though these concepts are very promising, two main disadvantages are still remaining. On the one hand, the high fabrication and integration cost and on the other hand the undesired adverse buckling of the suspended membranes. This thesis addresses these two problems and consists of two main parts: • PECVD dielectric material investigation and stress control resulting in membranes shape engineering. • Implementation and characterization of novel tunable optical devices with tailored shapes of the suspended membranes. For this purposes, low-cost PECVD technology is investigated and developed in detail. The macro- and microstress of silicon nitride and silicon dioxide are controlled over a wide range. Furthermore, the effect of stress on the optical and mechanical properties of the suspended membranes and on the microcavities is evaluated. Various membrane shapes (concave, convex and planar) with several radii of curvature are fabricated. Using this resonator shape engineering, microcavity devices such as non tunable and tunable Fabry-Pérot filters, VCSELs and PIN photodetectors are succesfully implemented. The fabricated Fabry-Pérot filters cover a spectral range of over 200nm and show resonance linewidths down to 1.5nm. By varying the stress distribution across the vertical direction within a DBR, the shape and the radius of curvature of the top membrane are explicitely tailored. By adjusting the incoming light beam waist to the curvature, the fundamental resonant mode is supported and the higher order ones are suppressed. For instance, a tunable VCSEL with 26 nm tuning range, 400µW maximal output power, 47nm free spectral range and over 57dB side mode suppresion ratio (SMSR) is demonstrated. Other technologies, such as introducing light emitting organic materials in microcavities are also investigated.

A critical account of the relationship between institutional trust, risk perception, and technology acceptance with an application to genetically modified foods

This article critically reflects on the widely held view of a causal chain with trust in public authorities impacting technology acceptance via perceived risk. It first puts forward conceptual reason against this view, as the presence of risk is a precondition for trust playing a role in decision making. Second, results from consumer surveys in Italy and Germany are presented that support the associationist model as counter hypothesis. In that view, trust and risk judgments are driven by and thus simply indicators of higher order attitudes toward a certain technology which determine acceptance instead. The implications of these findings are discussed.