Machine learning applied to the identification of underwater objects

This was a project supported by the Brazilian Navy, with the overall goal of employing machine learning to aid in the classification of static and underwater threats using the information gathered by multiple sensors. My contribution to the project was the implementation of an unsupervised neural network to capture the general behaviour of ULF electromagnetic fields received using ocean surface EM sensors, providing high-probability data points to reduce the number of processing data points to a significant data group while upsampling the sensor grid.