Roberto Célio Limão de Oliveira

New approach for T-wave end detection on electrocardiogram: Performance in noisy conditions

BackgroundThe detection of T-wave end points on electrocardiogram (ECG) is a basic procedure for ECG processing and analysis. Several methods have been proposed and tested, featuring high accuracy and percentages of correct detection. Nevertheless, their performance in noisy conditions remains an open problem.MethodsA new approach and algorithm for T-wave end location based on the computation of Trapeziums areas is proposed and validated (in terms of accuracy and repeatability), using signals from the Physionet QT Database. The performance of the proposed algorithm in noisy conditions has been tested and compared with one of the most used approaches for estimating the T-wave end point: the method based on the threshold on the first derivative.ResultsThe results indicated that the proposed approach based on Trapeziums areas outperformed the baseline method with respect to accuracy and repeatability. Also, the proposed method is more robust to wideband noise.ConclusionsThe trapezium-based approach has a good performance in noisy conditions and does not rely on any empirical threshold. It is very adequate for use in scenarios where the levels of broadband noise are significant.

Optimization of neural networks through grammatical evolution and a genetic algorithm

Neural codification confers scalability and search space reduction.The parallel genome scan engine increases the implicit parallelism of the GA.Our approach rewards economical ANNs which have better generalization capacity.Reduction in chromosome length from 512 to 180 bits.Our NEA outperforms other methods, providing the lowest computational effort. This paper proposes a hybrid neuro-evolutive algorithm (NEA) that uses a compact indirect encoding scheme (IES) for representing its genotypes (a set of ten production rules of a Lindenmayer System with memory), moreover has the ability to reuse the genotypes and automatically build modular, hierarchical and recurrent neural networks. A genetic algorithm (GA) evolves a Lindenmayer System (L-System) that is used to design the neural networks architecture. This basic neural codification confers scalability and search space reduction in relation to other methods. Furthermore, the system uses a parallel genome scan engine that increases both the implicit parallelism and convergence of the GA. The fitness function of the NEA rewards economical artificial neural networks (ANNs) that are easily implemented. The NEA was tested on five real-world classification datasets and three well-known datasets for time series forecasting (TSF). The results are statistically compared against established state-of-the-art algorithms and various forecasting methods (ADANN, ARIMA, UCM, and Forecast Pro). In most cases, our NEA outperformed the other methods, delivering the most accurate classification and time series forecasting with the least computational effort. These superior results are attributed to the improved effectiveness and efficiency of NEA in the decision-making process. The result is an optimized neural network architecture for solving classification problems and simulating dynamical systems.

Pricing analysis in the Brazilian energy market: A decision tree approach

There is a general consensus that electricity price forecasting is an important task nowadays since power market players are interested in the maximization of profit and the minimization of risk. In this context, this paper proposes the use of Data-Mining techniques to predict the short-term electricity price in the Brazilian market. In Brazil, the market model adopted has unique characteristics with a centralized form of dispatch due to the predominance of hydro generation. To apply the proposed prediction model, all features of the Brazilian electricity market are considered, such as the transmission restrictions among geo-electrical regions and the price dependency with storage energy in reservoirs. In the proposed prediction model, the electricity price is the dependent variable and the monthly time series data sets from the Brazilian system (such as power load, stored energy and thermal generation) are the independent variables. First, clustering of the data samples is performed to group similar behavior of the attributes. After that, a decision tree algorithm is applied to extract if-then rules from database. The rules obtained allow the identification of attributes that most influence the short-term electricity price. Results show that the proposed model can be an attractive tool to all electricity market players to forecast the short-term electricity price and mitigate the risks in purchasing power.

A New Cooperative Evolutionary Multi-Swarm Optimizer Algorithm Based on CUDA Architecture Applied to Engineering Optimization

This paper presents a new Cooperative Evolutionary Multi-Swarm Optimization Algorithm (CEMSO-GPU) based on CUDA parallel architecture applied to solve engineering problems. The focus of this approach is: the use of the concept of master/slave swarm with a mechanism of data sharing; and, the parallelism method based on the paradigm of General Purpose Computing on Graphics Processing Units (GPGPU) with CUDA architecture, brought by NVIDIA corporation. All these improvements were made aiming to produce better solutions in fewer iterations of the algorithm and to improve the search for best results. The algorithm was tested for some well-known engineering problems (WBD, ATD, MWTCS, SRD-11) and the results compared to other approaches.

Automatic design of Neural Networks with L-Systems and genetic algorithms - A biologically inspired methodology

In this paper we introduce a biologically plausible methodology capable to automatically generate Artificial Neural Networks (ANNs) with optimum number of neurons and adequate connection topology. In order to do this, three biological metaphors were used: Genetic Algorithms (GA), Lindenmayer Systems (L-Systems) and ANNs. The methodology tries to mimic the natural process of nervous system growing and evolution, using L-Systems as a recipe for development of the neurons and its connections and the GA to evolve and optimize the nervous system architecture suited for an specific task. The technique was tested on three well known simple problems, where recurrent networks topologies must be evolved. A more complex problem, involving time series learning was also proposed for application. The experiments results shows that our proposal is very promising and can generate appropriate neural networks architectures with an optimal number of neurons and connections, good generalization capacity, smaller error and large noise tolerance.

Genetic algorithms and treatment of multiple objectives in the allocation of capacitor banks in an electric power distribution system

Genetic Algorithm (GA) is a non-parametric optimization technique that is frequently used in problems of combinatory nature with discrete or continuous variables. Depending on the evaluation function used this optimization technique may be applied to solve problems containing more than one objective. In treating with multi-objective evaluation functions it is important to have an adequate methodology to solve the multiple objectives problem so that each partial objective composing the evaluation function is adequately treated in the overall optimal solution. In this paper the multi-objective optimization problem is treated in details and a typical example concerning the allocation of capacitor banks in a real distribution grid is presented. The allocation of capacitor banks corresponds to one of the most important problems related to the planning of electrical distribution networks. This problem consists of determining, with the smallest possible cost, the placement and the dimension of each capacitor bank to be installed in the electrical distribution grid with the additional objectives of minimizing the voltage deviations and power losses. As many other problems of planning electrical distribution networks, the allocation of capacitor banks is characterized by the high complexity in the search of the optimum solution. In this context, the GA comes as a viable tool to obtaining practical solutions to this problem. Simulation results obtained with a real electrical distribution grid are presented and demonstrate the effectiveness of the methodology used.

Game theory as a new paradigm for phenotype characterization of genetic algorithms

In this paper, it is presented a new way to characterize the phenotype in the context of Genetic Algorithms through the use of Game Theory as a theoretical foundation to define a new phase in the algorithm, named Social Interaction. It is executed before the reproduction phase and allows individuals to fight for their own survival improving their fitness according to the rules of a game. Thereby, a new algorithm is presented and some good results were produced for Traveling Salesman Problem an improvement in Genetic Algorithm execution.

Kalman Filter Embedded in FPGA to Improve Tracking Performance in Ballistic Rockets

This research project deals with the application of dynamic modeling methods for prediction and correction (filtering) of state variables in the calculation of the position and speed training rockets launched at Alcantara Launch Center. Initially, an approach is made of the means involved in launching and tracking of rockets, with the purpose of obtaining an understanding of the acquisition of radar signals, which are input of the filter. Then it made an approach of mathematical treatment of the filter dynamic model to obtain the equations used in the prediction and filtering of the rocket trajectory. Finally is made the application of Kalman filter algorithm in LabVIEW (Laboratory Virtual Instrumentation Engineering Workbench) FPGA (Field Programmable Gate Array) to estimate the position and velocity of the rocket. The main contribution of this work is to obtain a gain in processing velocity of Kalman filtering using parallelism at the hardware level, implementing a reconfigurable FPGA architecture, ensuring a platform fast enough for radars with high precision and good tracking capability of rockets.

Network Intrusion Detection System Using Data Mining

The aim of this study is to simulate a network traffic analyzer that is part of an Intrusion Detection System - IDS, the main focus of research is data mining and for this type of application the steps that precede the data mining : data preparation (possibly involving cleaning data, data transformations, selecting subsets of records, data normalization) are considered fundamental for a good performance of the classifiers during the data mining stage. In this context, this paper discusses and presents as a contribution not only the classifiers that were used in the problem of intrusion detection, but also the initial stage of data preparation. Therefore, we tested the performance of three classifiers on the KDDCUP’99 benchmark intrusion detection dataset and selected the best classifiers. We initially tested a Decision Tree and a Neural Network using this dataset, suggesting improvements by reducing the number of attributes from 42 to 27 considering only two classes of detection, normal and intrusion. Finally, we tested the Decision Tree and Bayesian Network classifiers considering five classes of attack: Normal, DOS, U2R, R2L and Probing. The experimental results proved that the algorithms used achieved high detection rates (DR) and significant reduction of false positives (FP) for different types of network intrusions using limited computational resources.

A multipopulation cultural algorithm based on genetic algorithm for the MKP

Genetic Algorithms (GAs) have been used to solve the NP-complete problems effectively such as the Multi Knapsack Problem (MKP). This work presents a combination between CAs and GAs with multipopulation model for the MKP. The benchmark simulation results indicate that the addition of multipopulation model improving optimization performance for CAs. To show the importance of multipopulation in the CAs, several MKP computational tests are performed for some benchmark problems