DeepMoleNet is a deep learning package for molecular properties prediction. It is developed with a multilevel attention mechanism to enable chemical interpretable insights being fused into multitask learning through (1) weighting contributions from various atoms and (2) taking the atom-centered symmetry functions (ACSFs) as the teacher descriptor, rather than using ACSFs as input in the conventional way. The properties including dipole moment, HOMO, and Gibbs free energy within chemical accuracy are achieved by using multiple benchmarks, both at the equilibrium and non-equilibrium geometries.
To cite this algorithm, please reference: Ziteng Liu, Liqiang Lin, Qingqing Jia, Zheng Cheng, Yanyan Jiang, Yanwen Guo*, Jing Ma*. "Transferable multilevel attention neural network for accurate prediction of quantum chemistry properties via multitask learning." J. Chem. Inf. Model. 2021, 61, 3, 1066–1082.
To request the code and data, please complete the LICENSE FORM, scan and email it to majing@nju.edu.cn.