Análise do comportamento de técnicas de correspondência binocular para imagens em alta definição

This work presents an analysis of the behavior of some algorithms usually available in stereo correspondence literature, with full HD images (1920x1080 pixels) to establish, within the precision dilemma versus runtime applications which these methods can be better used. The images are obtained by a system composed of a stereo camera coupled to a computer via a capture board. The OpenCV library is used for computer vision operations and processing images involved. The algorithms discussed are an overall method of search for matching blocks with the Sum of the Absolute Value of the difference (Sum of Absolute Differences - SAD), a global technique based on cutting energy graph cuts, and a so-called matching technique semi -global. The criteria for analysis are processing time, the consumption of heap memory and the mean absolute error of disparity maps generated.

Estimação e previsão no processo INARCH(2)

In the last decades the study of integer-valued time series has gained notoriety due to its broad applicability (modeling the number of car accidents in a given highway, or the number of people infected by a virus are two examples). One of the main interests of this area of study is to make forecasts, and for this reason it is very important to propose methods to make such forecasts, which consist of nonnegative integer values, due to the discrete nature of the data. In this work, we focus on the study and proposal of forecasts one, two and h steps ahead for integer-valued second-order autoregressive conditional heteroskedasticity processes [INARCH (2)], and in determining some theoretical properties of this model, such as the ordinary moments of its marginal distribution and the asymptotic distribution of its conditional least squares estimators. In addition, we study, via Monte Carlo simulation, the behavior of the estimators for the parameters of INARCH(2) processes obtained using three di erent methods (Yule- Walker, conditional least squares, and conditional maximum likelihood), in terms of mean squared error, mean absolute error and bias. We present some forecast proposals for INARCH(2) processes, which are compared again via Monte Carlo simulation. As an application of this proposed theory, we model a dataset related to the number of live male births of mothers living at Riachuelo city, in the state of Rio Grande do Norte, Brazil.

Estimação e previsão no processo INARCH(2)

In the last decades the study of integer-valued time series has gained notoriety due to its broad applicability (modeling the number of car accidents in a given highway, or the number of people infected by a virus are two examples). One of the main interests of this area of study is to make forecasts, and for this reason it is very important to propose methods to make such forecasts, which consist of nonnegative integer values, due to the discrete nature of the data. In this work, we focus on the study and proposal of forecasts one, two and h steps ahead for integer-valued second-order autoregressive conditional heteroskedasticity processes [INARCH (2)], and in determining some theoretical properties of this model, such as the ordinary moments of its marginal distribution and the asymptotic distribution of its conditional least squares estimators. In addition, we study, via Monte Carlo simulation, the behavior of the estimators for the parameters of INARCH(2) processes obtained using three di erent methods (Yule- Walker, conditional least squares, and conditional maximum likelihood), in terms of mean squared error, mean absolute error and bias. We present some forecast proposals for INARCH(2) processes, which are compared again via Monte Carlo simulation. As an application of this proposed theory, we model a dataset related to the number of live male births of mothers living at Riachuelo city, in the state of Rio Grande do Norte, Brazil.

Aplicação de técnicas de aprendizado de máquina no reconhecimento de classes estruturais de proteínas

Nowadays, classifying proteins in structural classes, which concerns the inference of patterns in their 3D conformation, is one of the most important open problems in Molecular Biology. The main reason for this is that the function of a protein is intrinsically related to its spatial conformation. However, such conformations are very difficult to be obtained experimentally in laboratory. Thus, this problem has drawn the attention of many researchers in Bioinformatics. Considering the great difference between the number of protein sequences already known and the number of three-dimensional structures determined experimentally, the demand of automated techniques for structural classification of proteins is very high. In this context, computational tools, especially Machine Learning (ML) techniques, have become essential to deal with this problem. In this work, ML techniques are used in the recognition of protein structural classes: Decision Trees, k-Nearest Neighbor, Naive Bayes, Support Vector Machine and Neural Networks. These methods have been chosen because they represent different paradigms of learning and have been widely used in the Bioinfornmatics literature. Aiming to obtain an improvment in the performance of these techniques (individual classifiers), homogeneous (Bagging and Boosting) and heterogeneous (Voting, Stacking and StackingC) multiclassification systems are used. Moreover, since the protein database used in this work presents the problem of imbalanced classes, artificial techniques for class balance (Undersampling Random, Tomek Links, CNN, NCL and OSS) are used to minimize such a problem. In order to evaluate the ML methods, a cross-validation procedure is applied, where the accuracy of the classifiers is measured using the mean of classification error rate, on independent test sets. These means are compared, two by two, by the hypothesis test aiming to evaluate if there is, statistically, a significant difference between them. With respect to the results obtained with the individual classifiers, Support Vector Machine presented the best accuracy. In terms of the multi-classification systems (homogeneous and heterogeneous), they showed, in general, a superior or similar performance when compared to the one achieved by the individual classifiers used - especially Boosting with Decision Tree and the StackingC with Linear Regression as meta classifier. The Voting method, despite of its simplicity, has shown to be adequate for solving the problem presented in this work. The techniques for class balance, on the other hand, have not produced a significant improvement in the global classification error. Nevertheless, the use of such techniques did improve the classification error for the minority class. In this context, the NCL technique has shown to be more appropriated

Influência da fadiga no comportamento neuromuscular do quadríceps femoral, após reconstrução do ligamento cruzado anterior

Muscle fatigue is a phenomenon that promotes physiological and biomechanical disorders and their changes in healthy subjects have been widely studied and have significant importance for care in preventing injuries, but we do not have many information about its effects in patients after ACL reconstruction. Thus, this study is to analyze the effects of fatigue on neuromuscular behavior of quadriceps after ACL reconstruction. To reach this objective, participants were forty men, twenty healthy (26,90 ± 6,29 years) and twenty after ACL reconstruction (29,75 ± 7,01 years) with a graft of semitendinosus and gracilis tendons, between four to six months after surgery. At first, there was an assessment of joint position sense (JPS) at the isokinetic dynamometer at a speed of 5°/s and target angle of 45° to analyze the absolute error of JPS. Next, we applied the a muscle fatigue protocol, running 100 repetitions of isokinetic knee flexion-extension at 90°/s. Concurrently with this protocol, there was the assessment of muscle performance, as the peak torque (PT) and fatigue index, and electromyographic activity (RMS and median frequency). Finally, we repeated the assessment of JPS. The statistical analysis showed that patients after ACL reconstruction have, even under normal conditions, the amended JPS compared with healthy subjects and that after fatigue, both have disturbances in the JPS, but this alteration is significantly exacerbated in patients after ACL reconstruction. About muscle performance, we could notice that these patients have a lower PT, although there are no differences between the dynamometric and EMG fatigue index. These findings show the necessity about the cares of pacients with ACL reconstruction in respect of the risks of articulate instability and overload in ligamentar graft

Atividade eletroencefalográfica de pacientes com acidente vascular cerebral na aprendizagem motora de um jogo baseado em realidade virtual

Stroke is the leading cause of long-term disability among adults and motor relearning is essential in motor sequelae recovery. Therefore, various techniques have been proposed to achieve this end, among them Virtual Reality. The aim of the study was to evaluate electroencephalographic activity of stroke patients in motor learning of a virtual reality-based game. The study included 10 patients with chronic stroke, right-hande; 5 with left brain injury (LP), mean age 48.8 years (± 4.76) and 5 with injury to the right (RP), mean age 52 years (± 10.93). Participants were evaluated for electroencephalographic (EEG) activity and performance while performing 15 repetitions of darts game in XBOX Kinect and also through the NIHSS, MMSE, Fugl-Meyer and the modified Ashworth scale. Patients underwent a trainning with 45 repetitions of virtual darts game, 12 sessions in four weeks. After training, patients underwent reassessment of EEG activity and performance in virtual game of darts (retention). Data were analyzed using ANOVA for repeated measures. According to the results, there were differences between the groups (PD and PE) in frequencies Low Alpha (p = 0.0001), High Alpha (p = 0.0001) and Beta (p = 0.0001). There was an increase in alpha activation powers and a decrease in beta in the phase retention of RP group. In LP group was observed increased alpha activation potency, but without decrease in beta activation. Considering the asymmetry score, RP group increased brain activation in the left hemisphere with the practice in the frontal areas, however, LP group had increased activation of the right hemisphere in fronto-central areas, temporal and parietal. As for performance, it was observed a decrease in absolute error in the game for RP group between assessment and retention (p = 0.015), but this difference was not observed for LP group (p = 0.135). It follows then that the right brain injury patients benefited more from darts game training in the virtual environment with respect to the motor learning process, reducing neural effort in ipsilesionais areas and errors with the practice of the task. In contrast, patients with lesions in left hemisphere decrease neural effort in contralesionais areas important for motor learning and showed no performance improvements with practice of 12 sessions of virtual dart game. Thus, the RV can be used in rehabilitation of stroke patients upper limb, but the laterality of the injury should be considered in programming the motor learning protocol.