Military Land Mines Classifier using K-Nearest Neighbor
Project Overview
This project focuses on classifying land mines using the K-Nearest Neighbor algorithm. It was my first machine learning project, where I developed a model to determine if a detected device is a mine or not based on magnetic anomaly data.
Project Description
The Land Mines Data Set is used to determine whether a land is safe to build and live on. There have been many approaches to obtain land mine detection, but so far there has not been a 100% success rate. To perform land mine detection, a process consisting of designing sensors, analyzing data and making decisions is carried out.
The magnetic anomaly method works by using the principle of calculating anomalies generated from objects from the magnetic field that disturb its structure, the magnetic field itself and the data taken to determine the movement and position of the land mine.
The determination of the parameters, namely position, depth and direction of movement using magnetic anomalies has been used since 1970. The Land Mines Data Set is multivariate in nature having 4 attributes. There are 3 independent variables from the input parameters, namely Voltage (V), High (H), and Soil Type (S), while for the output parameters, using the dependent variable, namely Mine Type (M).
Results
Achieved 100% accuracy on binary classification of landmines with only 50 training samples!
Learning curve showing 100% accuracy achieved with approximately 50 training samples
The learning curve demonstrates that our KNN model achieves perfect classification accuracy with a relatively small number of training samples. This indicates that the features extracted from the magnetic anomaly data provide strong signals for distinguishing between mines and non-mines.
Conclusion
This project demonstrates the effectiveness of the K-Nearest Neighbor algorithm for land mine detection using magnetic anomaly data. The model's ability to achieve 100% accuracy makes it a promising approach for real-world applications in land mine detection, potentially saving lives and making land safe for development.