Juan Contreras
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Juan Contreras has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
- Mariana Pino
Psychology Program, Universidad Autónoma del Caribe, Barranquilla, Atlántico, Colombia
2 shared publications - Aldo Lovo Ayala
Escuela Naval de Cadetes “Almirante Padilla”, Armada Nacional de Colombia, Cartagena 131001, Colombia
1 shared publication Alvaro Quintana
Escuela Naval Almirante Padilla, Corp. Univ. Rafael Nunez, Cartagena, Colombia
1 shared publication- Carlos Soto
Mechanical Engineering Program, Universidad Autónoma del Caribe, Barranquilla 080020, Colombia
1 shared publication - Daniela Ospino-Balcázar
ENS Research Group, ENS Group S.A.S.
1 shared publication - David Javier Muñoz Aldana
Universidad del Cauca. Escuela Naval de Cadetes Almirante Padilla
1 shared publication
Latest Publications
- Open Access
- 18 December 2025
- Published by MDPI AG in Computation
- Vol. 13 (12) , 296
Ship Model Identification Using Interpretable 4-DOF Maneuverability Models for River Combat Boat
Ship maneuverability models are typically defined by three degrees of freedom: surge, sway, and yaw. However, patrol vessels operating in riverine environments often exhibit significant roll motion during course changes, necessitating the inclusion of this dynamic. This study develops interpretable machine learning models capable of predicting vessel behavior in four degrees of freedom (4-DoF): surge, sway, yaw, and roll. A dataset of 125 h of simulated maneuvers was employed, including 29 h of out-of-distribution (OOD) conditions to test model generalization. Four models were implemented and compared over a 15-step prediction horizon: linear regression, third-order polynomial regression, a state-space model obtained via the N4SID algorithm, and an AutoRegressive model with eXogenous inputs (ARX). Results demonstrate that all models captured the essential vessel dynamics, with the state-space model achieving the best overall performance (e.g., NMSE = 0.0246 for surge velocity on test data and 0.0499 under OOD conditions). Variable-wise, surge and sway showed the lowest errors, roll rate remained stable, and yaw rate was the most sensitive to distribution shifts. Model-wise, the ARX model achieved the lowest NMSE for surge prediction (0.0149), while regression-based models provided interpretable yet less accurate alternatives. Multi-horizon evaluation (1-, 5-, 15-, and 30-step) under OOD conditions confirmed a consistent monotonic degradation across models. These findings validate the feasibility of using interpretable machine learning models for predictive control, autonomous navigation, and combat scenario simulation in riverine operations.
- 18 December 2025
- Published by MDPI AG in Computation
- Vol. 13 (12) , 296
Open AccessShip Model Identification Using Interpretable 4-DOF Maneuverability Models for River Combat Boat
Ship maneuverability models are typically defined by three degrees of freedom: surge, sway, and yaw. However, patrol vessels operating in riverine environments often exhibit significant roll motion during course changes, necessitating the inclusion of this dynamic. This study develops interpretable machine learning models capable of predicting vessel behavior in four degrees of freedom (4-DoF): surge, sway, yaw, and roll. A dataset of 125 h of simulated maneuvers was employed, including 29 h of out-of-distribution (OOD) conditions to test model generalization. Four models were implemented and compared over a 15-step prediction horizon: linear regression, third-order polynomial regression, a state-space model obtained via the N4SID algorithm, and an AutoRegressive model with eXogenous inputs (ARX). Results demonstrate that all models captured the essential vessel dynamics, with the state-space model achieving the best overall performance (e.g., NMSE = 0.0246 for surge velocity on test data and 0.0499 under OOD conditions). Variable-wise, surge and sway showed the lowest errors, roll rate remained stable, and yaw rate was the most sensitive to distribution shifts. Model-wise, the ARX model achieved the lowest NMSE for surge prediction (0.0149), while regression-based models provided interpretable yet less accurate alternatives. Multi-horizon evaluation (1-, 5-, 15-, and 30-step) under OOD conditions confirmed a consistent monotonic degradation across models. These findings validate the feasibility of using interpretable machine learning models for predictive control, autonomous navigation, and combat scenario simulation in riverine operations.
- Open Access
- 17 January 2024
- Published by Hipatia Press in International Journal of Educational Psychology
- Vol. 13 (1) , 1-18
Neuropsychological Assessment in Schooled Adolescent Offenders and Non-Offenders
Adolescents who break the law have experienced situations that increase the likelihood of becoming involved in criminal activities such as drug use, gang involvement, adverse economic conditions, among others. All this, added to their stage of human development, which is characterized by physical, cognitive, social and emotional changes, can lead them to have deficiencies in their cognitive processes and at the same time present educational difficulties. The purpose of this study is to evaluate different cognitive processes of these adolescents in comparison to a control group with similar characteristics but who have not committed any crime and whose education has not been interrupted. For this purpose, were included (n = 62) adolescent offenders and (n = 62) adolescent non-offenders of male sex and aged 14 to 18 years was taken. Basic sociodemographic data on their education and psychoactive substance use were collected, as well as cognitive data with tests such as Ineco Frontal Screening for executive functions, Montreal Cognitive Assessment for general functions, among others. The results showed significant differences in executive functions, attentional processes, memory and language. These difficulties can be key to school performance, therefore, educational interventions adapted to these adolescents are suggested.
- 17 January 2024
- Published by Hipatia Press in International Journal of Educational Psychology
- Vol. 13 (1) , 1-18
Open AccessNeuropsychological Assessment in Schooled Adolescent Offenders and Non-Offenders
Adolescents who break the law have experienced situations that increase the likelihood of becoming involved in criminal activities such as drug use, gang involvement, adverse economic conditions, among others. All this, added to their stage of human development, which is characterized by physical, cognitive, social and emotional changes, can lead them to have deficiencies in their cognitive processes and at the same time present educational difficulties. The purpose of this study is to evaluate different cognitive processes of these adolescents in comparison to a control group with similar characteristics but who have not committed any crime and whose education has not been interrupted. For this purpose, were included (n = 62) adolescent offenders and (n = 62) adolescent non-offenders of male sex and aged 14 to 18 years was taken. Basic sociodemographic data on their education and psychoactive substance use were collected, as well as cognitive data with tests such as Ineco Frontal Screening for executive functions, Montreal Cognitive Assessment for general functions, among others. The results showed significant differences in executive functions, attentional processes, memory and language. These difficulties can be key to school performance, therefore, educational interventions adapted to these adolescents are suggested.
- Open Access
Machine Learning Applications to Identify Young Offenders Using Data from Cognitive Function Tests
Machine learning techniques can be used to identify whether deficits in cognitive functions contribute to antisocial and aggressive behavior. This paper initially presents the results of tests conducted on delinquent and nondelinquent youths to assess their cognitive functions. The dataset extracted from these assessments, consisting of 37 predictor variables and one target, was used to train three algorithms which aim to predict whether the data correspond to those of a young offender or a nonoffending youth. Prior to this, statistical tests were conducted on the data to identify characteristics which exhibited significant differences in order to select the most relevant features and optimize the prediction results. Additionally, other feature selection methods, such as Boruta, RFE, and filter, were applied, and their effects on the accuracy of each of the three machine learning models used (SVM, RF, and KNN) were compared. In total, 80% of the data were utilized for training, while the remaining 20% were used for validation. The best result was achieved by the K-NN model, trained with 19 features selected by the Boruta method, followed by the SVM model, trained with 24 features selected by the filter method.
Open AccessMachine Learning Applications to Identify Young Offenders Using Data from Cognitive Function Tests
Machine learning techniques can be used to identify whether deficits in cognitive functions contribute to antisocial and aggressive behavior. This paper initially presents the results of tests conducted on delinquent and nondelinquent youths to assess their cognitive functions. The dataset extracted from these assessments, consisting of 37 predictor variables and one target, was used to train three algorithms which aim to predict whether the data correspond to those of a young offender or a nonoffending youth. Prior to this, statistical tests were conducted on the data to identify characteristics which exhibited significant differences in order to select the most relevant features and optimize the prediction results. Additionally, other feature selection methods, such as Boruta, RFE, and filter, were applied, and their effects on the accuracy of each of the three machine learning models used (SVM, RF, and KNN) were compared. In total, 80% of the data were utilized for training, while the remaining 20% were used for validation. The best result was achieved by the K-NN model, trained with 19 features selected by the Boruta method, followed by the SVM model, trained with 24 features selected by the filter method.
- 1 August 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Classification of small boats using fuzzy classifier
Drug trafficking organizations are using “go-fast” boats, small fishing boats and commercial containers to smuggle illegal drugs. Because that, detection, classification, and tracking of small vessels are important tasks for improving the security of complex maritime systems. This paper presents a simple method based on the least square method, averaging operator, instead of T-norm operator, and triangular membership function with 0.5 interpolations, in order to get a singleton fuzzy classifier from experimental data. The maxim number of fuzzy rules is limited to the sum of the number of membership function of each input variable; it means that the number of fuzzy rules is reduced with respect to classical fuzzy systems. An application for boats classification from their acoustic signatures is shown.
- 1 August 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Classification of small boats using fuzzy classifier
Drug trafficking organizations are using “go-fast” boats, small fishing boats and commercial containers to smuggle illegal drugs. Because that, detection, classification, and tracking of small vessels are important tasks for improving the security of complex maritime systems. This paper presents a simple method based on the least square method, averaging operator, instead of T-norm operator, and triangular membership function with 0.5 interpolations, in order to get a singleton fuzzy classifier from experimental data. The maxim number of fuzzy rules is limited to the sum of the number of membership function of each input variable; it means that the number of fuzzy rules is reduced with respect to classical fuzzy systems. An application for boats classification from their acoustic signatures is shown.
- 13 May 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Latin America Transactions
- Vol. 13 (4) , 1148-1152
INMO: A Web Architecture For Real Estate Search Systems
This paper describes the development of a real estate electronic marketplace based on platform J2EE. A novel real estate search method based on Google Maps APIs is included. We explain the approach we propose on a prototypical implementation exhibiting the described functionality.
- 13 May 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Latin America Transactions
- Vol. 13 (4) , 1148-1152
INMO: A Web Architecture For Real Estate Search Systems
This paper describes the development of a real estate electronic marketplace based on platform J2EE. A novel real estate search method based on Google Maps APIs is included. We explain the approach we propose on a prototypical implementation exhibiting the described functionality.
- 8 September 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Latin America Transactions
- Vol. 12 (6) , 1149-1154
Fuzzy Classifier for the Diagnosis of Pathology on the Vertebral Column
This paper describes the training and application of a fuzzy singleton classifier that is used for the diagnosis of pathologies of the vertebral column, discriminating patients as belonging to one out of three categories: Normal, Disk Hernia and Spondylolisthesis. Each input variable is partitioned into triangular membership functions so that consecutive fuzzy sets exhibit and specific overlapping of 0.5. Singleton consequents are employed and least square method is used to adjust the consequents. Applications are conducted on the vertebral column data set from University of California Irvine UCI Machine Learning Repository. Features of the dataset are obtained from sagittal panoramic radiographies of the spine. The dataset includes 310 instances (310 patients) having 6 real valued attributes in 3 classes
- 8 September 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Latin America Transactions
- Vol. 12 (6) , 1149-1154
Fuzzy Classifier for the Diagnosis of Pathology on the Vertebral Column
This paper describes the training and application of a fuzzy singleton classifier that is used for the diagnosis of pathologies of the vertebral column, discriminating patients as belonging to one out of three categories: Normal, Disk Hernia and Spondylolisthesis. Each input variable is partitioned into triangular membership functions so that consecutive fuzzy sets exhibit and specific overlapping of 0.5. Singleton consequents are employed and least square method is used to adjust the consequents. Applications are conducted on the vertebral column data set from University of California Irvine UCI Machine Learning Repository. Features of the dataset are obtained from sagittal panoramic radiographies of the spine. The dataset includes 310 instances (310 patients) having 6 real valued attributes in 3 classes
- 1 October 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Implementación de redes neuronales y lógica difusa para la clasificación de patrones obtenidos por un Sónar
This article describes a methodology using neural network models Adaline architecture and linguistically interpretable fuzzy systems, both algorithms were used to classify data signal rocks and metals, obtained through sonar. First, the neural network requires training match each input vector with the corresponding output vector for comparison with the desired output, and obtained, through feedback differential, an algorithm that minimizes the error. Secondly, the diffuse pattern contains overlapping of triangular sets to adjust the number of data sets for the antecedent, and singletons for to the consequent. In the evaluation of the rules are used instead of operators average T-norm and the consequents are adjust using recursive least squares. The promising aspect of this research was to achieve a good accuracy in the validation, after training of neural networks, and apply the fuzzy model without sacrificing the fuzzy system interpretability. Both methods are used making little modifications of parameters for obtain the best success percentage, the lowest mean square error (MSE), decreasing execution time, reducing effort in the processing machine and without recourse to other artificial intelligence techniques.
- 1 October 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Implementación de redes neuronales y lógica difusa para la clasificación de patrones obtenidos por un Sónar
This article describes a methodology using neural network models Adaline architecture and linguistically interpretable fuzzy systems, both algorithms were used to classify data signal rocks and metals, obtained through sonar. First, the neural network requires training match each input vector with the corresponding output vector for comparison with the desired output, and obtained, through feedback differential, an algorithm that minimizes the error. Secondly, the diffuse pattern contains overlapping of triangular sets to adjust the number of data sets for the antecedent, and singletons for to the consequent. In the evaluation of the rules are used instead of operators average T-norm and the consequents are adjust using recursive least squares. The promising aspect of this research was to achieve a good accuracy in the validation, after training of neural networks, and apply the fuzzy model without sacrificing the fuzzy system interpretability. Both methods are used making little modifications of parameters for obtain the best success percentage, the lowest mean square error (MSE), decreasing execution time, reducing effort in the processing machine and without recourse to other artificial intelligence techniques.
- 1 December 2012
- Published by Trans Tech Publications, Ltd. in Advanced Materials Research
- Vol. 629, 784-791
Generating Singleton Fuzzy Models from Data with Interpretable Partition
This paper presents a new methodology for obtaining singleton fuzzy model from experimental data. Each input variable is partitioned into triangular membership functions so that consecutive fuzzy sets exhibit and specific overlapping of 0.5. The recursive least squares method is employed to adjust the singleton consequences and the gradient descent method is employed to update only the modal value of each triangular membership function to preserve the overlap and reducing the number of parameters to be estimated. Applications to a function approximation problem and to a pattern classification problem are illustrated.
- 1 December 2012
- Published by Trans Tech Publications, Ltd. in Advanced Materials Research
- Vol. 629, 784-791
Generating Singleton Fuzzy Models from Data with Interpretable Partition
This paper presents a new methodology for obtaining singleton fuzzy model from experimental data. Each input variable is partitioned into triangular membership functions so that consecutive fuzzy sets exhibit and specific overlapping of 0.5. The recursive least squares method is employed to adjust the singleton consequences and the gradient descent method is employed to update only the modal value of each triangular membership function to preserve the overlap and reducing the number of parameters to be estimated. Applications to a function approximation problem and to a pattern classification problem are illustrated.
- 1 March 2011
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Generating fuzzy controllers for ship steering
In this paper we present a new method to generate fuzzy controller from training data. The antecedent partition uses triangular sets with 0.5 interpolations avoiding the presence of complex overlapping that happens in other methods. Singleton consequents are employed and least square method is used to adjust the consequents. This approach is not a hybrid system and does not employ other techniques, like neural network or genetic algorithm. The applicability of the proposed approach is demonstrated by application for controlling the directional heading of a cargo ship.
- 1 March 2011
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Generating fuzzy controllers for ship steering
In this paper we present a new method to generate fuzzy controller from training data. The antecedent partition uses triangular sets with 0.5 interpolations avoiding the presence of complex overlapping that happens in other methods. Singleton consequents are employed and least square method is used to adjust the consequents. This approach is not a hybrid system and does not employ other techniques, like neural network or genetic algorithm. The applicability of the proposed approach is demonstrated by application for controlling the directional heading of a cargo ship.