Artificial intelligence applied to nuclear forensics via deep learning
Nuclear forensics aims to investigate the origin and history of nuclear or radioactive materials via analytical techniques. Using microscopy images, we have performed deep learning analysis to assess multiple calcination conditions and processing routes of nuclear samples. We have applied, compared and fine-tuned via transfer learning state-of-the-art convolutional neural networks, performing various classification and regression tasks. In addition to applying standard neural network architectures, we have investigated the use of parallel networks for multi-magnification acquisitions, and quantified model uncertainty during inference.