A machine leaning-based approach is proposed for the inverse analysis of the anisotropy parameters of solid–liquid interfacial free energy. The interface shape distribution (ISD) map, which characterizes the details of the dendrite morphology, was selected as the input of a convolutional neural network (CNN). The ISD maps for a free-growing dendrite during the isothermal solidification of a model alloy system were obtained by quantitative phase-field simulations and used as the training and test data for the CNN. Two anisotropy parameters were estimated with errors of less than 5%, which can be further improved by increasing the size of the training dataset.