Explicit Inductive Bias for Transfer Learning with Convolutional Networks
Abstract
Regularization methods that explicitly keep a fine-tuned model close to its pre-trained initialization improve inductive transfer learning, and a simple L2 penalty to the pre-trained model is recommended as a baseline.
In inductive transfer learning, fine-tuning pre-trained convolutional networks substantially outperforms training from scratch. When using fine-tuning, the underlying assumption is that the pre-trained model extracts generic features, which are at least partially relevant for solving the target task, but would be difficult to extract from the limited amount of data available on the target task. However, besides the initialization with the pre-trained model and the early stopping, there is no mechanism in fine-tuning for retaining the features learned on the source task. In this paper, we investigate several regularization schemes that explicitly promote the similarity of the final solution with the initial model. We show the benefit of having an explicit inductive bias towards the initial model, and we eventually recommend a simple L^2 penalty with the pre-trained model being a reference as the baseline of penalty for transfer learning tasks.
Get this paper in your agent:
hf papers read 1802.01483 Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash Models citing this paper 0
No model linking this paper
Datasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper
Collections including this paper 0
No Collection including this paper