AdaFilter: Adaptive Filter Fine-tuning for Deep Transfer Learning
Preprint
Abstract
There is an increasing number of pre-trained deep neural network models. However, it is still unclear how to effectively use these models for a new task. Transfer learning, which aims to transfer knowledge from source tasks to a target task, is an effective solution to this problem. Fine-tuning is a popular transfer learning technique for deep neural networks where a few rounds of training are applied to the parameters of a pre-trained model to adapt them to a new task. Despite its popularity, in this paper, we show that fine-tuning suffers from several drawbacks. We propose an adaptive fine-tuning approach, called AdaFilter, which selects only a part of the convolutional filters in the pre-trained model to optimize on a per-example basis. We use a recurrent gated network to selectively fine-tune convolutional filters based on the activations of the previous layer. We experiment with 7 public image classification datasets and the results show that AdaFilter can reduce the average classification error of the standard fine-tuning by 2.54%.Keywords
All Related Versions
This publication has 8 references indexed in Scilit:
- Depthwise Convolution Is All You Need for Learning Multiple Visual DomainsProceedings of the AAAI Conference on Artificial Intelligence, 2019
- Continual lifelong learning with neural networks: A reviewNeural Networks, 2019
- Borrowing Treasures from the Wealthy: Deep Transfer Learning through Selective Joint Fine-Tuning2017
- Question Answering through Transfer Learning from Large Fine-grained Supervision Data2017
- Dynamic Image Networks for Action Recognition2016
- Human-level concept learning through probabilistic program inductionScience, 2015
- Geodesic flow kernel for unsupervised domain adaptation2012
- A Survey on Transfer LearningIEEE Transactions on Knowledge and Data Engineering, 2009