| About this Abstract |
| Meeting |
2019 TMS Annual Meeting & Exhibition
|
| Symposium
|
Characterization of Materials through High Resolution Imaging
|
| Presentation Title |
Learning CDI Reconstructions with Backpropagation |
| Author(s) |
Youssef Nashed |
| On-Site Speaker (Planned) |
Youssef Nashed |
| Abstract Scope |
Synchrotron radiation light source facilities are leading the way to ultrahigh resolution X-ray imaging. High resolution imaging is essential to understanding the fundamental structure and interaction of materials at the smallest length scale possible. Coherent diffraction imaging (CDI) achieves nanoscale imaging by replacing traditional objective lenses by pixelated area detectors and computational image reconstruction. We present our work for solving CDI reconstruction problems through fitting a physics based model to measured data. The model parameters are learned in a similar manner to deep neural networks, utilizing the backpropagation method as implemented in Google TensorFlow package. This approach has advantages in terms of speed and accuracy compared to current state of the art algorithms, and demonstrates re-purposing the deep learning backpropagation algorithm to solve general inverse problems that are prevalent in materials imaging research. |
| Proceedings Inclusion? |
Planned: Supplemental Proceedings volume |