A Multi-Category Inverse Design Neural Network and Its Application to Diblock Copolymers

In this work, we design a multi-category inverse neural network to map ordered periodic structures physical parameters. The model consists of two parts, classifier and Structure-Parameter-Mapping (SPM) subnets. is used identify structures, the SPM subnets are predict parameters for desired structures. We also present an extensible reciprocal-space data augmentation method guarantee rotation translation invariant apply proposed two-dimensional diblock copolymers based on Landau–Brazovskii model. Results show that has high accuracy in predicting Moreover, idea multi-categorization can be extended other problems.