Abstract
The vast amount of design freedom in disordered systems expands the parameter space for signal processing. However, this large degree of freedom has hindered the deterministic design of disordered systems for target functionalities. Here, we employ a machine learning approach for predicting and designing wave-matter interactions in disordered structures, thereby identifying scale-free properties for waves. To abstract and map the features of wave behaviors and disordered structures, we develop disorder-to-localization and localization-to-disorder convolutional neural networks, each of which enables the instantaneous prediction of wave localization in disordered structures and the instantaneous generation of disordered structures from given localizations. We demonstrate that the structural properties of the network architectures lead to the identification of scale-free disordered structures having heavy-tailed distributions, thus achieving multiple orders of magnitude improvement in robustness to accidental defects. Our results verify the critical role of neural network structures in determining machine-learning-generated real-space structures and their defect immunity.
Highlights
The vast amount of design freedom in disordered systems expands the parameter space for signal processing
To substitute the time-consuming and problem-specific process of calculating analytical microstructural descriptors while making full use of microstructural information, we can envisage the use of multiple-layer neural network (NN) models as datadriven descriptors to identify the relationship between disordered structures and wave behaviors
Because the ML-generated lattice deformation is strongly related to the weights of the output neurons in the L2D convolutional neural networks (CNNs), the apparent stochastic difference between normal-random seed structures and scale-free L2D CNN outputs raises an interesting open question; the training process of deep NNs could inherently possess the scale-free property
Summary
The vast amount of design freedom in disordered systems expands the parameter space for signal processing. To substitute the time-consuming and problem-specific process of calculating analytical microstructural descriptors while making full use of microstructural information, we can envisage the use of multiple-layer neural network (NN) models as datadriven descriptors to identify the relationship between disordered structures and wave behaviors. This deep-learning-based framework[21,22], one of the powerful machine-learning (ML). Tools, has proven successful for abstracting the features of data sets in pattern recognition, decision making, and language translation[23,24] when carefully preprocessed data can be used
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
