The virtual sample generation and screening methods proposed in this paper can effectively solve the modeling problem of small samples. The reliable process parameters can be obtained by optimizing the model trained by the semi-synthetic samples.
Keyphrases