FPFrank Jesus Valderrama Purizaca

Frank Jesus Valderrama Purizaca

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Frank Jesus Valderrama Purizaca has a total of 4 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Desarrollo de un modelo predictivo de las propiedades mecánicas del suelo usando redes neuronales artificiales

    by Frank Jesus Valderrama Purizaca, Sócrates Pedro Muñoz Pérez, Victor A. Tuesta-Monteza, Heber Ivan Mejía-Cabrera

    Determinar las propiedades del suelo es una tarea necesaria pero costosa en el diseño de pavimentos, por dicha razón, en este estudio se desarrollaron cuatro redes neuronales artificiales (RNA) basadas en perceptrón multicapa para predecir la máxima densidad seca (MDD), el óptimo contenido de humedad (OMC), el valor de la relación de soporte de California (CBR) al 95% de la MDD y el CBR al 100% de la MDD, respectivamente. El método considera un dataset con 285 ejemplos, definición de arquitectura base mediante optimización bayesiana y validación cruzada, modificación de la arquitecturay los hiperparámetros para mejorar el desempeño. Las RNA se entrenaron considerando 3000 épocas, función ReLU,tasa de aprendizaje, dropout; fueron evaluadas con el coeficiente de correlación (R) y el error cuadrático medio (MSE) y predijeron la MDD con R=0,90, OMC con R=0,87, CBR al 95% con R=0,92, CBR al 100% con R=0,89, respectivamente, demostrando que los modelos son eficientes para predecir las propiedades del suelo.

  • Open Access

    Importance of artificial neural networks in civil engineering: a systematic review of the literature

    by Frank Jesús Valderrama Purizaca, Daniel Armando Chávez Barturen, Sócrates Pedro Muñoz Pérez, Victor A. Tuesta-Monteza, Heber Ivan Mejía-Cabrera

    Artificial neural networks (ANN) have a relevant role nowadays; several areas apply this technique due to the advantages they have to solve complex problems with many constraints compared to traditional methods, which are becoming outdated. Very little is known about this technique and its application in different branches of Civil Engineering. For this reason, the present research aims to conduct a systematic review of the literature to identify the use of this technique and to determine the results of the application of ANN models in civil engineering. A total of 41 scientific articles were included, distributed as follows: 6 in Scopus, 1 in ScienceDirect, 23 in ProQuest, 7 in Google Scholar, 2 in DialNet, 2 in SciELO. It was found that ANNs are used to predict or forecast variables associated with the fields of study in civil engineering; 8 applications of ANN were found for concrete properties, 11 for soil properties, 5 for seismic analysis, 9 for hydraulics, 7 for real estate valuation and 1 for bridge design. Likewise, it was found that the multilayer Perceptron is the most used ANN model, achieving an average R2 of 0.99, which shows advantages to solve problems with precision, in shorter times, with missing data in the data sets, as well as the reduction of the error factor.