Artificial Neural Networks for Determining Magnetospheric Conditions

Abstract: This chapter presents a neural-network-based technique that allows for the reconstruction of the global, time-varying distribution of some physical quantity Q, that has been sparsely sampled at various locations within the magnetosphere, and at different times. We begin with a general introduction to the problem of prediction and specification, and why it is important and difficult to achieve with existing methods. We then provide a basic introduction to neural networks, and describe our tech…