Identifying buried objects using the neural network approach

We present a neural network approach for detecting conducting anomalies, e.g., underground buried objects and those located in sedimentary layers in seafloors. The first step in this approach is to simulate the scattering from buried objects by using a suitable computational modeling tool. The electric and magnetic field values, thus computed, are then used as inputs to a neural network and the associated conductivities are treated as targets. The neural network is trained to associate the conductivity profiles with the computed field values. Finally, we demonstrate that a trained neural network can be used to estimate the conductivities and sizes of new objects, not originally employed to train the network.