Inteligência artificial aplicada na previsão de mercados financeiros

Neste documento, são investigados vários métodos usados na inteligência artificial, com o objetivo de obter previsões precisas da evolução dos mercados financeiros. O uso de ferramentas lineares como os modelos AR, MA, ARMA e GARCH têm muitas limitações, pois torna-se muito difícil adaptá-los às não linearidades dos fenómenos que ocorrem nos mercados. Pelas razões anteriormente referidas, os algoritmos como as redes neuronais dinâmicas (TDNN, NARX e ESN), mostram uma maior capacidade de adaptação a estas não linearidades, pois não fazem qualquer pressuposto sobre as distribuições de probabilidade que caracterizam estes mercados. O facto destas redes neuronais serem dinâmicas, faz com que estas exibam um desempenho superior em relação às redes neuronais estáticas, ou outros algoritmos que não possuem qualquer tipo de memória. Apesar das vantagens reveladas pelas redes neuronais, estas são um sistema do tipo black box, o que torna muito difícil extrair informação dos pesos da rede. Isto significa que estes algoritmos devem ser usados com precaução, pois podem tornar-se instáveis.

Biologically motivated model for object detection and identification in real-world scenes

The classical computer vision methods can only weakly emulate some of the multi-level parallelisms in signal processing and information sharing that takes place in different parts of the primates’ visual system thus enabling it to accomplish many diverse functions of visual perception. One of the main functions of the primates’ vision is to detect and recognise objects in natural scenes despite all the linear and non-linear variations of the objects and their environment. The superior performance of the primates’ visual system compared to what machine vision systems have been able to achieve to date, motivates scientists and researchers to further explore this area in pursuit of more efficient vision systems inspired by natural models. In this paper building blocks for a hierarchical efficient object recognition model are proposed. Incorporating the attention-based processing would lead to a system that will process the visual data in a non-linear way focusing only on the regions of interest and hence reducing the time to achieve real-time performance. Further, it is suggested to modify the visual cortex model for recognizing objects by adding non-linearities in the ventral path consistent with earlier discoveries as reported by researchers in the neuro-physiology of vision.

Novel actuator design to overcome stiction and backlash for low-velocity applications

This paper presents a novel actuator design that ameliorates or eliminates the effects of non-linearities that are characteristically present in geared actuator systems and which are very problematic for low velocity applications. The design centres on the providing an internal rotational element within a single actuator to ensure operation of actuator away from the stiction region, whilst allowing zero velocity external output of the actuator. The construction has the added advantage of substantially reducing backlash. The prototype comprises two commercially available servo-actuators to test the principle of operation and results presented indicate that the concept is worth exploring further.

A neural network enhanced generalized minimum variance self-tuning proportional, integral and derivative control algorithm for complex dynamic systems

A neural network enhanced proportional, integral and derivative (PID) controller is presented that combines the attributes of neural network learning with a generalized minimum-variance self-tuning control (STC) strategy. The neuro PID controller is structured with plant model identification and PID parameter tuning. The plants to be controlled are approximated by an equivalent model composed of a simple linear submodel to approximate plant dynamics around operating points, plus an error agent to accommodate the errors induced by linear submodel inaccuracy due to non-linearities and other complexities. A generalized recursive least-squares algorithm is used to identify the linear submodel, and a layered neural network is used to detect the error agent in which the weights are updated on the basis of the error between the plant output and the output from the linear submodel. The procedure for controller design is based on the equivalent model, and therefore the error agent is naturally functioned within the control law. In this way the controller can deal not only with a wide range of linear dynamic plants but also with those complex plants characterized by severe non-linearity, uncertainties and non-minimum phase behaviours. Two simulation studies are provided to demonstrate the effectiveness of the controller design procedure.

Intergenerational persistence in health in developing countries : the penalty of gender inequality?

This paper is motivated to investigate the often neglected payoff to investments in the health of girls and women in terms of next generation outcomes. This paper investigates the intergenerational persistence of health across time and region as well as across the distribution of maternal health. It uses comparable microdata on as many as 2.24 million children born of about 0.6 million mothers in 38 developing countries in the 31 year period, 1970–2000. Mother's health is indicated by her height, BMI and anemia status. Child health is indicated by mortality risk and anthropometric failure. We find a positive relationship between maternal and child health across indicators and highlight non-linearities in these relationships. The results suggest that both contemporary and childhood health of the mother matter and that the benefits to the next generation are likely to be persistent. Averaging across the sample, persistence shows a considerable decline over time. Disaggregation shows that the decline is only significant in Latin America. Persistence has remained largely constant in Asia and has risen in Africa. The paper provides the first cross-country estimates of the intergenerational persistence in health and the first estimates of trends.

The ocean's gravitational potential energy budget in a coupled climate model

This study examines, in a unified fashion, the budgets of ocean gravitational potential energy (GPE) and available gravitational potential energy (AGPE) in the control simulation of the coupled atmosphere–ocean general circulation model HadCM3. Only AGPE can be converted into kinetic energy by adiabatic processes. Diapycnal mixing supplies GPE, but not AGPE, whereas the reverse is true of the combined effect of surface buoyancy forcing and convection. Mixing and buoyancy forcing, thus, play complementary roles in sustaining the large scale circulation. However, the largest globally integrated source of GPE is resolved advection (+0.57 TW) and the largest sink is through parameterized eddy transports (-0.82 TW). The effect of these adiabatic processes on AGPE is identical to their effect on GPE, except for perturbations to both budgets due to numerical leakage exacerbated by non-linearities in the equation of state.

Evaluating forecasts from SETAR models of exchange rates

We consider the forecasting performance of two SETAR exchange rate models proposed by Kräger and Kugler [J. Int. Money Fin. 12 (1993) 195]. Assuming that the models are good approximations to the data generating process, we show that whether the non-linearities inherent in the data can be exploited to forecast better than a random walk depends on both how forecast accuracy is assessed and on the ‘state of nature’. Evaluation based on traditional measures, such as (root) mean squared forecast errors, may mask the superiority of the non-linear models. Generalized impulse response functions are also calculated as a means of portraying the asymmetric response to shocks implied by such models.

The influence of ARIMA-GARCH parameters in feed forward neural networks prediction

The objective of this article is to find out the influence of the parameters of the ARIMA-GARCH models in the prediction of artificial neural networks (ANN) of the feed forward type, trained with the Levenberg-Marquardt algorithm, through Monte Carlo simulations. The paper presents a study of the relationship between ANN performance and ARIMA-GARCH model parameters, i.e. the fact that depending on the stationarity and other parameters of the time series, the ANN structure should be selected differently. Neural networks have been widely used to predict time series and their capacity for dealing with non-linearities is a normally outstanding advantage. However, the values of the parameters of the models of generalized autoregressive conditional heteroscedasticity have an influence on ANN prediction performance. The combination of the values of the GARCH parameters with the ARIMA autoregressive terms also implies in ANN performance variation. Combining the parameters of the ARIMA-GARCH models and changing the ANN`s topologies, we used the Theil inequality coefficient to measure the prediction of the feed forward ANN.

A aleatoriedade do passeio na BOVESPA: testando a eficiência do mercado acionário brasileira

Este artigo investiga versões do modelo de passeio aleatório dos preços de ativos em diversos horizontes de tempo, para carteiras diversificadas de ações no mercado brasileiro. Evidências contrárias a tal modelo são observadas nos horizontes diário e semanal, caracterizados por persistência. As evidências são mais fracas em períodos mais recentes. Encontramos também sazonalidades diárias, incluindo o efeito segunda-feira, e mensais. Adicionalmente, um padrão de assimetria de autocorrelações cruzadas de primeira ordem entre os retornos de carteiras de firmas agrupadas segundo seu tamanho também é observado, indicando no caso de retornos diários e semanais que retornos de firmas grandes ajudam a prever retornos de firmas pequenas. Evidências de não linearidades nos retornos são observadas em diversos horizontes de tempo.

Controle adaptativo robusto para um modelo desacoplado de um robô móvel

This thesis presents a new structure of robust adaptive controller applied to mobile robots (surface mobile robot) with nonholonomic constraints. It acts in the dynamics and kinematics of the robot, and it is split in two distinct parts. The first part controls the robot dynamics, using variable structure model reference adaptive controllers. The second part controls the robot kinematics, using a position controller, whose objective is to make the robot to reach any point in the cartesian plan. The kinematic controller is based only on information about the robot configuration. A decoupling method is adopted to transform the linear model of the mobile robot, a multiple-input multiple-output system, into two decoupled single-input single-output systems, thus reducing the complexity of designing the controller for the mobile robot. After that, a variable structure model reference adaptive controller is applied to each one of the resulting systems. One of such controllers will be responsible for the robot position and the other for the leading angle, using reference signals generated by the position controller. To validate the proposed structure, some simulated and experimental results using differential drive mobile robots of a robot soccer kit are presented. The simulator uses the main characteristics of real physical system as noise and non-linearities such as deadzone and saturation. The experimental results were obtained through an C++ program applied to the robot soccer kit of Microrobot team at the LACI/UFRN. The simulated and experimental results are presented and discussed at the end of the text

Antimony glasses with large nonlinear refractive indices, small two-photon absorption coefficients and ultrafast response at telecom wavelengths

Nonlinear (NL) optical properties of antimony oxide based glasses (AG) were characterized for excitation wavelengths from 800 to 1600 m. The NL refractive indices, n2, and the two-photon absorption (TPA) coefficient, β, have been evaluated using the Z-scan technique. Values of n2≈ 10-15 - 10-14 cm2/W of electronic origin were measured and negligible TPA coefficients (β < 0.003 cm/GW) were determined. The response time of the nonlinearity is faster than 100 fs as determined using the Kerr shutter technique. The figure-of-merit usually considered for all-optical switching, T = 2βλ/n2 , indicates that AG are very good materials for ultrafast switches at telecom wavelengths. © 2007 IEEE.

Shunt active compensation based on the Conservative Power Theory current's decomposition

Considering the operation of shunt active compensators, such as active power filters, this paper proposes possible compensation strategies by means of the recent formulation of the Conservative Power Theory (CPT). The CPT current's decomposition results in several current components, which are associated with specific load characteristics (power transfer, energy storage, unbalances and/or non linearities). These current components are used for the definition of different compensation strategies, which can be selective in terms of minimizing particular disturbing effects. In order to validate the applicability of these new compensation strategies, simulation and experimental results for three-phase four-wire systems are presented. © 2011 IEEE.

Fenômenos não-lineares, incluindo-se os não-ideias, em captura de energia utilizando-se dispositivos piezoelétricos

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Metodologia de projeto cooperativo para condicionadores de qualidade de energia

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Consideração da ação de dispositivos facts em um método automático de análise de estabilidade de sistemas de energia elétrica

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)