Scalable Extraction of Information from Spatiotemporal Patterns of Chemoresponsive Liquid Crystals Using Topological Descriptors

Chemoresponsive liquid crystals (LCs) can be engineeredto generateinformation-rich optical responses (in the form of space-timecolor and brightness patterns) when they are exposed to target gascontaminants. We investigate the use of topological descriptors (Eulercharacteristic, lacunarity, and fractal dimension) for extractingdifferent types of information from these complex optical responsesand show that these tools can enable the design of sensors and helpgain insights into the physical phenomena governing sensor responses.We provide a holistic perspective of topological descriptors usingthe theory of Minkowski functionals and fractal analysis, which allowsus to understand specific information that each descriptor extracts.We also show how to use the topological descriptors in conjunctionwith space-time filtration operations and color representationsto enrich the information extracted. We demonstrate how these capabilitiescan be used in flexible ways to facilitate unsupervised machine learning(ML) tasks (clustering and visualization) and supervised tasks (regressionand classification). We demonstrate the developments using real, high-throughputexperimental data sets for functionalized LC films that are exposedto different gaseous environments. We show that the topological descriptorsencode significant information and can be used to detect outliersin high-throughput data and visualize the temporal evolution of structure.Moreover, we show that the topological descriptors can be used topredict contaminant concentrations using simple ML models such assupport vector machines; notably, these ML models can achieve accuraciescomparable to those of powerful convolutional neural networks butwith a much lower computational cost (from hours to seconds) and usingless sophisticated computing hardware (CPUs instead of GPUs). Thisscalability enables the analysis of space-time data at highresolutions.