Fig. 1. Data-driven applications in the field of LC materials

Fig. 2. Self-assembled structures and phase transition simulations in LC systems：（a）LC molecule composed of five particles，（b）LC molecule composed of ten particles，（c）Initial structure of the LCS under random arrangement；Influence of binary liquid crystal molecule ratio on phase transition temperature under machine learning：（d，e）Prediction results from different machine learning methods and feature importance ranking under RF^［39］；（f）Accuracy，loss，and confusion function of the Inception network applied to the complex classification task of isotropic，nematic，cholesteric，and smectic textures^［41］；（g）Average accuracy and confusion matrix of the Inception model on the ChSm test set^［43］.

Fig. 3. Overview of machine learning algorithm for optimizing threshold voltage parameters of co-doped ZnO liquid crystals^［51］

Fig. 4. （a）Deep learning framework analyzes liquid crystal sensor images and calculates the bright area coverage ratio to detect different concentrations of Cd^2+［58］，hierarchical convolutional neural network for surfactant classification；（b）The first level classifies droplet patterns using grayscale micrographs；（c）The second level labels the surfactant type and concentration for bipolar liquid crystal droplets；（d）Performance evaluation through the confusion matrix from cross-validation^［60］.