Exploring diamondlike lattice thermal conductivity crystals via feature-based transfer learning

Ultrahigh lattice thermal conductivity materials hold great importance since they play a critical role in the management of electronic and optical devices. Models using machine learning can search for with outstanding higher-order properties like conductivity. However, lack sufficient data to train model is serious hurdle. Herein we show that big complement small accurate predictions when lower-order feature available are selected properly applied transfer learning. The connection between crystal information directly built neural network by transferring descriptors acquired through pre-trained property. Successful shows ability extrapolative prediction reveals anharmonicity. Transfer employed screen over 60000 compounds identify novel crystals serve as alternatives diamond.