In this work, a Reinforcement Learning (RL) method for a geospatial analysis of clandestine graves in Mexico is presented. For law enforcement authorities and families looking for missing people, the issue of clandestine graves and the discovery of remains is a complicated and sensitive one. The goal of the study is to assess area reduction exploration and visualize common trajectories to find clandestine graves using RL. The work uses various techniques to increase the effectiveness of the RL model, including estimating state values, action values, and optimal policies and adding function approximators.
Trajectory example
The final report with full documentation of the project is found at report/finalreport.pdf the trajectories which cannot be seen from the pdf are located under the report folder as well. All the code to reproduce the development can be found at the code folder.
This repository has the following structure:
├── README.md
├── code/
├── data/
├──── /generated_data/
├──── /raw_data/
├── images/
├──── /results/
└── report/
an archive folder is inside the code directory containing draft codes that are stored for future development processes