Deep into The Domain Shift: Transfer Learning through Dependence Regularization
Published in SSRN, 2022
In this papaer, we develop a novel domain adaptation method that can flexibly model the correspondence strength between source distributions and target distributions. We successfully apply it into two financial big-data scenarios: Hong Kong Exchange stock price prediction and JD.com customer credit defaults prediction. Large-scale experiments and ablation studies have demonstrated the effectiveness of our proposed models.