Generalization Bounds for Deep Transfer Learning Using Majority Predictor Accuracy
This work addresses the need for theoretical understanding and practical tools in transfer learning for machine learning researchers, though it appears incremental in nature.
The paper tackles the problem of analyzing generalization bounds for deep transfer learning models by introducing a quantity called majority predictor accuracy, which can be computed efficiently from data. The result shows that this quantity can serve as a practical transferability measure, as validated by experiments.
We analyze new generalization bounds for deep learning models trained by transfer learning from a source to a target task. Our bounds utilize a quantity called the majority predictor accuracy, which can be computed efficiently from data. We show that our theory is useful in practice since it implies that the majority predictor accuracy can be used as a transferability measure, a fact that is also validated by our experiments.