Evaluación de algoritmos de aprendizaje y método Wenner en tomografía eléctrica para la detección automática de fosas simuladas
Loading...
Share
Date
2021-06-08
Authors
Director(s)
Publisher
Universidad Antonio Nariño
Campus
Degree obtained
Document type
COAR type
http://purl.org/coar/resource_type/c_7a1f
Citation
Bibliographic Managers
Document Viewer
Select a file to preview:
item.page.resume
Abstract
Forced disappearance in Colombia constitutes a crime against humanity that violates human rights and undermines the well-being of families. In the country more than 80,500 people have been victims of this fact that only generates anxiety, uncertainty and fear. Although the Congress of the Republic approved Law 1408 to redeem the victims and relatives of the forced disappearance, in addition, to contribute in the search for the disappeared; the results have not been effective. The Justice and Peace Investigative Group of the Criminal Investigation Directorate has failed to implement the techniques that are time-consuming, expensive, and not very successful. In this sense and in order to contribute to the mitigation of the suffering of the victims, this experimental research work framed in forensic geophysics, presents the implementation of the Wenner Method and the validation of three Machine Learning models. From the models, machine learning algorithms such as Bagged Trees were implemented; Boostd Trees, Rusboosted, and Subspace KNN, which were applied with and without tags. In each of the analyzes, Matlab classification learning tools were used, in some cases, the data was processed by searching and filling in outliers or normalized values. However, it is the Bageed Trees model that had the highest accuracy and performance rates. According to the confusion matrix, the percentages ranged from 59 to 75.8% The results of the investigation become an ideal tool to contribute to the automatic detection of simulated graves in the country and a hope for thousands of relatives who still live with pain and absence in their hearts.