| About this Abstract |
| Meeting |
2019 TMS Annual Meeting & Exhibition
|
| Symposium
|
Characterization of Materials through High Resolution Imaging
|
| Presentation Title |
Deep Neural Networks for Feature Extraction and Image Reconstruction from Coherent X-ray Diffraction Imaging Data |
| Author(s) |
Mathew Cherukara, Youssef Nashed, Ross Harder |
| On-Site Speaker (Planned) |
Mathew Cherukara |
| Abstract Scope |
Coherent X-ray diffraction imaging (CDI) is a powerful technique for operando characterization. Visualizing defects, dynamics, and structural evolution using CDI, however, remains a grand challenge since state-of-the-art iterative reconstruction algorithms for CDI data are time-consuming and computationally expensive, which precludes real-time feedback. Furthermore, the reconstruction algorithms require human inputs to guide their convergence, which is a very subjective process. I will show results from training a neural network (NN) to classify and identify the defect structure of raw X-ray diffraction data without the need for any reconstruction through iterative phase retrieval. While the NN was trained on atomistic data, it performs at ~95% accuracy on real-world X-ray diffraction data, indicating the robustness of the network. I will also describe the development of BCDI NN, a deep generative network that can predict structure and phase from raw diffraction data in a fraction of a second on a standard desktop machine. |
| Proceedings Inclusion? |
Planned: Supplemental Proceedings volume |