91°µÍø

SHARE
Research Highlight

Deep Learning Infers Structures of General Lyotropic Phases from Small Angle Neutron Scattering

Deep Learning Infers Structures of General Lyotropic Phases from Small Angle Neutron Scattering
A convolutional neural network framework (middle panel), framework for analyzing small-angle scattering data from general lyotropic phases. The convolutional neural network architecture blends autoencoder elements with probabilistic latent variables to facilitate the inversion of real-space structures (left panel) derived from experimentally collected small-angle scattering data (right panel).

Scientific Achievement

A novel strategy that integrates convolutional neural networks and stochastically regulated matter wave field into a regression analysis framework enables inferring real-space structures of general lyotropic phases from small angle neutron scattering.

Significance and Impact

The study shows that deep learning overcomes the limitations of traditional analytical modeling in characterizing complex lyotropic phases, including intermediates with fused topological features.

Research Details

  • A regression analysis for small angle scattering data combining convolutional neural networks and regulated matter wave field.
  • The feasibility was tested using computational benchmarks and explicit small angle scattering demonstrations.

Work performed at the Center for Nanophase Materials Sciences and the Spallation Neutron Source

Chi-Huan Tung, Yu-Jung Hsiao, Hsin-Lung Chen, Guan-Rong Huang, Lionel Porcar, Ming-Ching Chang, Jan-Michael Carrillo, Yangyang Wang, Bobby G. Sumpter, Yuya Shinohara, Jon Taylor, Changwoo Do, and Wei-Ren Chen, Journal of Colloid and Interface Science 659,739–750 (2024).