Josiel Maimone de Figueiredo

Audio parameterization with robust frame selection for improved bird identification

Audio parameterization method with robust frame selection.Automated acoustic recognition of 40 bird species.HMM-based bird identification. A major challenge in the automated acoustic recognition of bird species is the audio segmentation, which aims to select portions of audio that contain meaningful sound events and eliminates segments that contain predominantly background noise or sound events of other origin. Here we report on the development of an audio parameterization method with integrated robust frame selection that makes use of morphological filtering applied on the spectrogram seen as an image. The morphological filtering allows to exclude from further processing certain audio events, which otherwise could cause misclassification errors. The Mel Frequency Cepstral Coefficients (MFCCs) computed for the selected audio frames offer a good representation of the spectral information for dominant vocalizations because the morphological filtering eliminates short bursts of noise and suppresses weak competing signals. Experimental validation of the proposed method on the identification of 40 bird species from Brazil demonstrated superior accuracy and faster operation than three traditional and recent approaches. This is expressed as reduction of the relative error rate by 3.4% and the overall operational time by 7.5% when compared to the second best result. The improved frame selection robustness, precision, and operational speed facilitate applications like multi-species identification of real-field recordings.

Study of How the Integration of Artificial Neural Network and Genetic Algorithm Should Be Made for Modeling Meteorological Data

Artificial Neural Networks (ANN) have been widely used to model several types of data. The precision of ANN models is dependent upon their configuration, i.e., input parameters, training algorithm and architecture configurations. The problem lies in the amount of possible combinations of these parameters which results in countless unique ANNs. One method of finding a good combination of ANN parameters is to use a Genetic Algorithm (GA). Several studies combine a GA with an ANN to solve problems, however, it is not clear which parameters of an ANN the GA should determine. This work performed thousands of tests to verify the best combinations of parameters to use in integrations between GA and ANN especially in modeling meteorological data. Results have shown that the best approach is to use GA to define the input variables, activation function and the number of neurons of the ANN. Other tests showed that this same combination had similar results with different types of data indicating that this work can perhaps be applied to several types of problems.

A framework for automating the configuration of OpenCL

Environmental research and scientific simulations use information acquired by sensors to validate the modeling and representation of environmental behaviors. The computational processing cost of this context tends to be extremely high due to the amount of information and the models calculation complexities which demand the use of computational parallel solutions. This paper presents JSeriesCL, a framework for parallel processing of spatiotemporal series using graphics processors (GPGPU), more specifically OpenCL. GPU is cheaper than other solutions for parallel processing, such as clusters or grid, and JSeriesCL changes the way that GPU are used because it automates the configuration and management aspects of such devices. Fractal dimension and SEBS were used to validate the application of JSeriesCL over environmental data. Our framework automates the overall management and configuration aspects of GPU.All source codes which use the framework are portable over distincts GPU.Programming errors and software maintenance are minimized.The maximum GPU process power is achieved with the framework.

SADCoRH A Dynamic System to Solve Conflicts in the Use of Water through the Generation of Rules by Genetic Algorithms

The sustainability of the water uses in the future depends on the planning and the rational management of water resources. As many governments around the world, Brazil has endorsed in 1997 the federal law 9.433 on water resources, which previews the development of management tools to support decision making on how to use and regulate the rights of rational water use. This work presents a computational tool, based on genetics algorithms, able to optimize water use allocation. The system is validated in a case study in the Cuiaba river basin, in Mato Grosso State, Brazil.