Xinran Ma and Z. C. Tu from Beijing Normal University and Shi-Ju Ran from Capital Normal University applied convolutional neural network (CNN) to predict the physical parameters of interacting Hamiltonians given the quantum many body wave functions at the ground states. QubismNet, the proposed method has two parts: 1. visualize the ground state wavefunctions / density matrices as images; 2. use CNN to map the images to the target Hamiltonian parameters. QubismNet is able to predict parameters beyond the training region such as predicting the magnetic fields near the critical point while the training states are far away from the critical point. This method may be helpful to infer the Hamiltonians given the designed ground state wavefunctions. Reference: arXiv:2012.03019