Quantum circuit learning to predict experimental chemical properties

We introduce quantum circuit learning (QCL) as an emerging regression algorithm for chemo- and materials-informatics. The supervised model, functioning on the rule of quantum mechanics, can learn various linear and non-linear functions from small datasets (< 100 records). Its prediction process is studied by analyzing and visualizing the complex latent variables. We also show that the QCL model was powerful in predicting values in extrapolating regions. The advantage is demonstrated with several experimental chemical databases. The superior accuracies against conventional models (random forest, support vector machine, and linear algorithms) promise a new possibility of exploring new materials, breaking the traditional performance limits.