Frozen Overparameterization: A Double Descent Perspective on Transfer Learning of Deep Neural Networks
This work addresses the generalization behavior in transfer learning for deep neural networks, providing insights into performance optimization, but it is incremental as it builds on existing overparameterization and double descent theories.
The paper investigates how transfer learning of deep neural networks generalizes, focusing on the double descent phenomenon and overparameterization, and finds that factors like dataset sizes, number of frozen layers, and task similarity can significantly affect test error and training dynamics, with experiments showing specific effects such as slower training from larger source datasets.
We study the generalization behavior of transfer learning of deep neural networks (DNNs). We adopt the overparameterization perspective -- featuring interpolation of the training data (i.e., approximately zero train error) and the double descent phenomenon -- to explain the delicate effect of the transfer learning setting on generalization performance. We study how the generalization behavior of transfer learning is affected by the dataset size in the source and target tasks, the number of transferred layers that are kept frozen in the target DNN training, and the similarity between the source and target tasks. We show that the test error evolution during the target DNN training has a more significant double descent effect when the target training dataset is sufficiently large. In addition, a larger source training dataset can yield a slower target DNN training. Moreover, we demonstrate that the number of frozen layers can determine whether the transfer learning is effectively underparameterized or overparameterized and, in turn, this may induce a freezing-wise double descent phenomenon that determines the relative success or failure of learning. Also, we show that the double descent phenomenon may make a transfer from a less related source task better than a transfer from a more related source task. We establish our results using image classification experiments with the ResNet, DenseNet and the vision transformer (ViT) architectures.