RESUMO
Randomization-based neural networks have gained wide acceptance in the scientific community owing to the simplicity of their algorithm and generalization capabilities. Random vector functional link (RVFL) networks and their variants are a class of randomization-based neural networks. RVFL networks have shown promising results in classification, regression, and clustering problems. For real-world applications, learning algorithms that can train with new samples over previous results are necessary because of to the constant generation of problems related to large-scale datasets. Various online sequential algorithms, commonly involving an initial learning phase followed by a sequential learning phase, have been proposed to address this issue. This paper presents a training algorithm based on multiple online sequential random vector functional link (OS-RVFL) networks for large-scale databases using a shared memory architecture. The training dataset is distributed among p OS-RVFL networks, which are trained in parallel using p threads. Subsequently, the test dataset samples are classified using each trained OS-RVFL network. Finally, a frequency criterion is applied to the results obtained from each OS-RVFL network to determine the final classification. Additionally, an equation was derived to reasonably predict the total training time of the proposed algorithm based on the learning time in the initial phase and the time scaling factor compared to the sequential learning phase. The results demonstrate a drastic reduction in training time because of data distribution and an improvement in accuracy because of the adoption of the frequency criterion.
RESUMO
Introduction: The analysis of injuries caused by traffic from a physical and mathematical perspective can help improve road safety strategies. Objective: Predict the dynamics of traffic fatalities in the states of Maryland and Massachusetts for the years 2004 and 2014 in the context of probabilistic random walk. Methods: An analysis was made of the number of total fatalities caused by traffic per year, in the states of Maryland and Massachusetts between the years 1994-2003 and 1994-2013. The behavior of these values was analyzed as a probabilistic random walk; for this, the probabilistic lengths were found for each year, during the period studied and four probability spaces were analyzed, with which it was possible to analyze their behavior, to establish a prediction of the number of total fatalities caused by traffic for the years 2004 and 2014. Results: The predictions for the years 2014 and 2004 for Maryland and Massachusetts when compared with the real values, the percentage of success was 98%. Main conclusion: The predictions for the years 2014 and 2004 for Maryland and Massachusetts when compared with the real values, the percentage of success was 98%. Conclusions: the behavior of traffic fatalities in Maryland and Massachusetts presented a predictable self-organization from the context of probabilistic random walk, constituting a useful tool for analyzing the operation of road safety strategies.
Antecedentes: El análisis de los accidentes de tránsito desde una perspectiva física y matemática puede ayudar a mejorar las estrategias viales de seguridad. Objetivo: Obtener una predicción de la dinámica de fatalidades a causa del tráfico en los estados de Maryland y Massachusetts para los años 2004 y 2014 en el contexto de la caminata al azar probabilista. Métodos: Se realizó un análisis del número de fatalidades totales causadas por el tráfico al año, en los estados de Maryland y Massachusetts entre los años 1994-2003 y 1994-2013. El comportamiento de estos valores fue analizado como una caminata al azar probabilista; para ello se hallaron las longitudes probabilistas para cada año, durante el periodo estudiado y se analizaron cuatro espacios de probabilidad, con los que fue posible analizar su comportamiento, para establecer una predicción del número de fatalidades totales causadas por el tráfico para los años 2004 y 2014. Resultados: Las predicciones para los años 2014 y 2004 para Maryland y Massachusetts al ser comparados con los valores reales el porcentaje de acierto fue del 98%. Conclusión principal: el comportamiento de las fatalidades de tráfico en Maryland y Massachusetts presentó una autoorganización predecible desde el contexto de la caminata al azar probabilista, constituyéndose como una herramienta útil para el análisis del funcionamiento de las estrategias de seguridad vial.