Aprendizado não-supervisionado em redes neurais pulsadas de base radial: um estudo da capacidade de agrupamento para a classificação de pixels

Redes neurais pulsadas - redes que utilizam uma codificação temporal da informação - têm despontado como uma promissora abordagem dentro do paradigma conexionista, emergente da ciência cognitiva. Um desses novos modelos é a rede neural pulsada com função de base radial, que é capaz de armazenar informação nos tempos de atraso axonais dos neurônios. Um algoritmo de aprendizado foi aplicado com sucesso nesta rede pulsada, que se mostrou capaz de mapear uma seqüência de pulsos de entrada em uma seqüência de pulsos de saída. Mais recentemente, um método baseado no uso de campos receptivos gaussianos foi proposto para codificar dados constantes em uma seqüência de pulsos temporais. Este método tornou possível a essa rede lidar com dados computacionais. O processo de aprendizado desta nova rede não se encontra plenamente compreendido e investigações mais profundas são necessárias para situar este modelo dentro do contexto do aprendizado de máquinas e também para estabelecer as habilidades e limitações desta rede. Este trabalho apresenta uma investigação desse novo classificador e um estudo de sua capacidade de agrupar dados em três dimensões, particularmente procurando estabelecer seus domínios de aplicação e horizontes no campo da visão computacional.

Tuning of fuzzy inference systems through unconstrained optimization techniques

This paper presents a new methodology for the adjustment of fuzzy inference systems. A novel approach, which uses unconstrained optimization techniques, is developed in order to adjust the free parameters of the fuzzy inference system, such as its intrinsic parameters of the membership function and the weights of the inference rules. This methodology is interesting, not only for the results presented and obtained through computer simulations, but also for its generality concerning to the kind of fuzzy inference system used. Therefore, this methodology is expandable either to the Mandani architecture or also to that suggested by Takagi-Sugeno. The validation of the presented methodology is accomplished through an estimation of time series. More specifically, the Mackey-Glass chaotic time series estimation is used for the validation of the proposed methodology.

Analysis of Structural Integrity Using an ARTMAP-Fuzzy Artificial Neural Network

This paper presents the application of artificial neural networks in the analysis of the structural integrity of a building. The main objective is to apply an artificial neural network based on adaptive resonance theory, called ARTMAP-Fuzzy neural network and apply it to the identification and characterization of structural failure. This methodology can help professionals in the inspection of structures, to identify and characterize flaws in order to conduct preventative maintenance to ensure the integrity of the structure and decision-making. In order to validate the methodology was modeled a building of two walk, and from this model were simulated various situations (base-line condition and improper conditions), resulting in a database of signs, which were used as input data for ARTMAP-Fuzzy network. The results show efficiency, robustness and accuracy.

Enhancing weak signal transmission through a feedforward network

The ability to transmit and amplify weak signals is fundamental to signal processing of artificial devices in engineering. Using a multilayer feedforward network of coupled double-well oscillators as well as Fitzhugh-Nagumo oscillators, we here investigate the conditions under which a weak signal received by the first layer can be transmitted through the network with or without amplitude attenuation. We find that the coupling strength and the nodes' states of the first layer act as two-state switches, which determine whether the transmission is significantly enhanced or exponentially decreased. We hope this finding is useful for designing artificial signal amplifiers.

Stochastic competitive learning in complex networks

Competitive learning is an important machine learning approach which is widely employed in artificial neural networks. In this paper, we present a rigorous definition of a new type of competitive learning scheme realized on large-scale networks. The model consists of several particles walking within the network and competing with each other to occupy as many nodes as possible, while attempting to reject intruder particles. The particle's walking rule is composed of a stochastic combination of random and preferential movements. The model has been applied to solve community detection and data clustering problems. Computer simulations reveal that the proposed technique presents high precision of community and cluster detections, as well as low computational complexity. Moreover, we have developed an efficient method for estimating the most likely number of clusters by using an evaluator index that monitors the information generated by the competition process itself. We hope this paper will provide an alternative way to the study of competitive learning.

Network-based stochastic semisupervised learning

Semisupervised learning is a machine learning approach that is able to employ both labeled and unlabeled samples in the training process. In this paper, we propose a semisupervised data classification model based on a combined random-preferential walk of particles in a network (graph) constructed from the input dataset. The particles of the same class cooperate among themselves, while the particles of different classes compete with each other to propagate class labels to the whole network. A rigorous model definition is provided via a nonlinear stochastic dynamical system and a mathematical analysis of its behavior is carried out. A numerical validation presented in this paper confirms the theoretical predictions. An interesting feature brought by the competitive-cooperative mechanism is that the proposed model can achieve good classification rates while exhibiting low computational complexity order in comparison to other network-based semisupervised algorithms. Computer simulations conducted on synthetic and real-world datasets reveal the effectiveness of the model.

Dynamics of Spatial Game Theory Networks with Novel Modifications

Introduction: Advances in biotechnology have shed light on many biological processes. In biological networks, nodes are used to represent the function of individual entities within a system and have historically been studied in isolation. Network structure adds edges that enable communication between nodes. An emerging fieldis to combine node function and network structure to yield network function. One of the most complex networks known in biology is the neural network within the brain. Modeling neural function will require an understanding of networks, dynamics, andneurophysiology. It is with this work that modeling techniques will be developed to work at this complex intersection. Methods: Spatial game theory was developed by Nowak in the context of modeling evolutionary dynamics, or the way in which species evolve over time. Spatial game theory offers a two dimensional view of analyzingthe state of neighbors and updating based on the surroundings. Our work builds upon this foundation by studying evolutionary game theory networks with respect to neural networks. This novel concept is that neurons may adopt a particular strategy that will allow propagation of information. The strategy may therefore act as the mechanism for gating. Furthermore, the strategy of a neuron, as in a real brain, isimpacted by the strategy of its neighbors. The techniques of spatial game theory already established by Nowak are repeated to explain two basic cases and validate the implementation of code. Two novel modifications are introduced in Chapters 3 and 4 that build on this network and may reflect neural networks. Results: The introduction of two novel modifications, mutation and rewiring, in large parametricstudies resulted in dynamics that had an intermediate amount of nodes firing at any given time. Further, even small mutation rates result in different dynamics more representative of the ideal state hypothesized. Conclusions: In both modificationsto Nowak's model, the results demonstrate the network does not become locked into a particular global state of passing all information or blocking all information. It is hypothesized that normal brain function occurs within this intermediate range and that a number of diseases are the result of moving outside of this range.

Curvature-driven smoothing: a learning algorithm for feed-forward networks

The performance of feed-forward neural networks in real applications can be often be improved significantly if use is made of a-priori information. For interpolation problems this prior knowledge frequently includes smoothness requirements on the network mapping, and can be imposed by the addition to the error function of suitable regularization terms. The new error function, however, now depends on the derivatives of the network mapping, and so the standard back-propagation algorithm cannot be applied. In this paper, we derive a computationally efficient learning algorithm, for a feed-forward network of arbitrary topology, which can be used to minimize the new error function. Networks having a single hidden layer, for which the learning algorithm simplifies, are treated as a special case.

A Bayesian perspective on stochastic neurocontrol

Control design for stochastic uncertain nonlinear systems is traditionally based on minimizing the expected value of a suitably chosen loss function. Moreover, most control methods usually assume the certainty equivalence principle to simplify the problem and make it computationally tractable. We offer an improved probabilistic framework which is not constrained by these previous assumptions, and provides a more natural framework for incorporating and dealing with uncertainty. The focus of this paper is on developing this framework to obtain an optimal control law strategy using a fully probabilistic approach for information extraction from process data, which does not require detailed knowledge of system dynamics. Moreover, the proposed control method framework allows handling the problem of input-dependent noise. A basic paradigm is proposed and the resulting algorithm is discussed. The proposed probabilistic control method is for the general nonlinear class of discrete-time systems. It is demonstrated theoretically on the affine class. A nonlinear simulation example is also provided to validate theoretical development.

Study of the Application of Neural Networks in Internet Traffic Engineering

In this study, we showed various approachs implemented in Artificial Neural Networks for network resources management and Internet congestion control. Through a training process, Neural Networks can determine nonlinear relationships in a data set by associating the corresponding outputs to input patterns. Therefore, the application of these networks to Traffic Engineering can help achieve its general objective: “intelligent” agents or systems capable of adapting dataflow according to available resources. In this article, we analyze the opportunity and feasibility to apply Artificial Neural Networks to a number of tasks related to Traffic Engineering. In previous sections, we present the basics of each one of these disciplines, which are associated to Artificial Intelligence and Computer Networks respectively.

Cloud condensation nuclei in pristine tropical rainforest air of Amazonia: size-resolved measurements and modeling of atmospheric aerosol composition and CCN activity

Atmospheric aerosol particles serving as cloud condensation nuclei (CCN) are key elements of the hydrological cycle and climate. We have measured and characterized CCN at water vapor supersaturations in the range of S=0.10-0.82% in pristine tropical rainforest air during the AMAZE-08 campaign in central Amazonia. The effective hygroscopicity parameters describing the influence of chemical composition on the CCN activity of aerosol particles varied in the range of kappa approximate to 0.1-0.4 (0.16+/-0.06 arithmetic mean and standard deviation). The overall median value of kappa approximate to 0.15 was by a factor of two lower than the values typically observed for continental aerosols in other regions of the world. Aitken mode particles were less hygroscopic than accumulation mode particles (kappa approximate to 0.1 at D approximate to 50 nm; kappa approximate to 0.2 at D approximate to 200 nm), which is in agreement with earlier hygroscopicity tandem differential mobility analyzer (H-TDMA) studies. The CCN measurement results are consistent with aerosol mass spectrometry (AMS) data, showing that the organic mass fraction (f(org)) was on average as high as similar to 90% in the Aitken mode (D <= 100 nm) and decreased with increasing particle diameter in the accumulation mode (similar to 80% at D approximate to 200 nm). The kappa values exhibited a negative linear correlation with f(org) (R(2)=0.81), and extrapolation yielded the following effective hygroscopicity parameters for organic and inorganic particle components: kappa(org)approximate to 0.1 which can be regarded as the effective hygroscopicity of biogenic secondary organic aerosol (SOA) and kappa(inorg)approximate to 0.6 which is characteristic for ammonium sulfate and related salts. Both the size dependence and the temporal variability of effective particle hygroscopicity could be parameterized as a function of AMS-based organic and inorganic mass fractions (kappa(p)=kappa(org) x f(org)+kappa(inorg) x f(inorg)). The CCN number concentrations predicted with kappa(p) were in fair agreement with the measurement results (similar to 20% average deviation). The median CCN number concentrations at S=0.1-0.82% ranged from N(CCN,0.10)approximate to 35 cm(-3) to N(CCN,0.82)approximate to 160 cm(-3), the median concentration of aerosol particles larger than 30 nm was N(CN,30)approximate to 200 cm(-3), and the corresponding integral CCN efficiencies were in the range of N(CCN,0.10/NCN,30)approximate to 0.1 to N(CCN,0.82/NCN,30)approximate to 0.8. Although the number concentrations and hygroscopicity parameters were much lower in pristine rainforest air, the integral CCN efficiencies observed were similar to those in highly polluted megacity air. Moreover, model calculations of N(CCN,S) assuming an approximate global average value of kappa approximate to 0.3 for continental aerosols led to systematic overpredictions, but the average deviations exceeded similar to 50% only at low water vapor supersaturation (0.1%) and low particle number concentrations (<= 100 cm(-3)). Model calculations assuming aconstant aerosol size distribution led to higher average deviations at all investigated levels of supersaturation: similar to 60% for the campaign average distribution and similar to 1600% for a generic remote continental size distribution. These findings confirm earlier studies suggesting that aerosol particle number and size are the major predictors for the variability of the CCN concentration in continental boundary layer air, followed by particle composition and hygroscopicity as relatively minor modulators. Depending on the required and applicable level of detail, the information and parameterizations presented in this paper should enable efficient description of the CCN properties of pristine tropical rainforest aerosols of Amazonia in detailed process models as well as in large-scale atmospheric and climate models.

A novel procedure for establishing clinching parameters for high strength steel sheet

The building industry has a particular interest in using clinching as a joining method for frame constructions of light-frame housing. Normally many clinch joints are required in joining of frames.In order to maximise the strength of the complete assembly, each clinch joint must be as sound as possible. Experimental testing is the main means of optimising a particular clinch joint. This includes shear strength testing and visual observation of joint cross-sections. The manufacturers of clinching equipment normally perform such experimental trials. Finite element analysis can also be used to optimise the tool geometry and the process parameter, X, which represents the thickness of the base of the joint. However, such procedures require dedicated software, a skilled operator, and test specimens in order to verify the finite element model. In addition, when using current technology several hours' computing time may be necessary. The objective of the study was to develop a simple calculation procedure for rapidly establishing an optimum value for the parameter X for a given tool combination. It should be possible to use the procedure on a daily basis, without stringent demands on the skill of the operator or the equipment. It is also desirable that the procedure would significantly decrease thenumber of shear strength tests required for verification. The experimental workinvolved tests in order to obtain an understanding of the behaviour of the sheets during clinching. The most notable observation concerned the stage of the process in which the upper sheet was initially bent, after which the deformation mechanism changed to shearing and elongation. The amount of deformation was measured relative to the original location of the upper sheet, and characterised as the C-measure. By understanding in detail the behaviour of the upper sheet, it waspossible to estimate a bending line function for the surface of the upper sheet. A procedure was developed, which makes it possible to estimate the process parameter X for each tool combination with a fixed die. The procedure is based on equating the volume of material on the punch side with the volume of the die. Detailed information concerning the behaviour of material on the punch side is required, assuming that the volume of die does not change during the process. The procedure was applied to shear strength testing of a sample material. The sample material was continuously hot-dip zinc-coated high-strength constructional steel,with a nominal thickness of 1.0 mm. The minimum Rp0.2 proof stress was 637 N/mm2. Such material has not yet been used extensively in light-frame housing, and little has been published on clinching of the material. The performance of the material is therefore of particular interest. Companies that use clinching on a daily basis stand to gain the greatest benefit from the procedure. By understanding the behaviour of sheets in different cases, it is possible to use data at an early stage for adjusting and optimising the process. In particular, the functionality of common tools can be increased since it is possible to characterise the complete range of existing tools. The study increases and broadens the amount ofbasic information concerning the clinching process. New approaches and points of view are presented and used for generating new knowledge.

An artificial neural network model for prediction of quality characteristics of apples during convective dehydration

In this study, the effects of hot-air drying conditions on color, water holding capacity, and total phenolic content of dried apple were investigated using artificial neural network as an intelligent modeling system. After that, a genetic algorithm was used to optimize the drying conditions. Apples were dried at different temperatures (40, 60, and 80 °C) and at three air flow-rates (0.5, 1, and 1.5 m/s). Applying the leave-one-out cross validation methodology, simulated and experimental data were in good agreement presenting an error < 2.4 %. Quality index optimal values were found at 62.9 °C and 1.0 m/s using genetic algorithm.

Forced-air, vacuum, and hydro precooling of cauliflower (Brassica oleracea L. var. botrytis cv. Freemont): part I. determination of precooling parameters

The aim of the present study was to precool cauliflower using forced-air, vacuum and high and low flow hydro cooling methods. The weight of the precooled cauliflower heads (5000±5 g) was measured before they were placed in standard plastic crates. Cauliflower heads, whose initial temperature was 23.5 ± 0.5 ºC, were cooled until the temperature reached at 1 ºC. During the precooling process, time-dependent temperature and energy consumption were measured, and during vacuum precooling, the decreasing pressure values were recorded, and a curve of time-dependent pressure decrease (vacuum) was built. The most suitable cooling method to precool cauliflower in terms of cooling time and energy consumption was vacuum, followed by the high and low flow hydro and forced-air precooling methods, respectively. The highest weight loss was observed in the vacuum precooling method, followed by the forced-air method. However, there was an increase in the weight of the cauliflower heads in the high and low flow hydro precooling method. The best colour and hardness values were found in the vacuum precooling method. Among all methods tested, the most suitable method to precool cauliflower in terms of cooling and quality parameters was the vacuum precooling method.

Deposition and photo-induced electrical resistivity of dip-coated NiO thin films from a precipitation process

Thin films of the semiconductor NiO are deposited using a straightforward combination of simple and versatile techniques: the co-precipitation in aqueous media along with the dip- coating process. The obtained material is characterized by gravimetric/differential thermal analysis (TG-DTA) and X-ray diffraction technique. TG curve shows 30 % of total mass loss, whereas DTA indicates the formation of the NiO phase about 578 K (305 C). X-ray diffraction (XRD) data confirms the FCC crystalline phase of NiO, whose crystallinity increases with thermal annealing temperature. UV-Vis optical absorption measurements are carried out for films deposited on quartz substrate in order to avoid the masking of bandgap evaluation by substrate spectra overlapping. The evaluated bandgap is about 3.0 eV. Current-voltage (I-V) curves measured for different temperatures as well as the temperature-dependent resistivity data show typical semiconductor behavior with the resistivity increasing with the decreasing of temperature. The Arrhenius plot reveals a level 233 meV above the conduction band top, which was attributed to Ni2+ vacancy level, responsible for the p-type electrical nature of NiO, even in undoped samples. Light irradiation on the films leads to a remarkable behavior, because above bandgap light induced a resistivity increase, despite the electron-hole generation. This performance was associated with excitation of the Ni 2+ vacancy level, due to the proximity between energy levels. © 2012 Springer Science+Business Media New York.