Signal integrity analysis of chip package based on machine learning

A machine learning approach to chip package signal integrity analysis is proposed. Combining the active learning model and the migrate learning model, the paper improves the performance of the simulation algorithm by fine-tuning the pretraining model, analyzes the convergent speed of the different scale models and the epoch needed before and after the freezing of the different learning rates, it is concluded that the convergence rate is the fastest when the labeled data is 80%, and the period of the unfrozen model is shorter than that of the frozen model.