Please use this identifier to cite or link to this item: https://doi.org/10.1109/WACV45572.2020.9093343
Title: GradMix: Multi-source transfer across domains and tasks
Authors: Li, J
Xu, Z
Wang, Y 
Zhao, Q
Kankanhalli, MS 
Keywords: cs.CV
cs.CV
Issue Date: 1-Mar-2020
Publisher: IEEE
Citation: Li, J, Xu, Z, Wang, Y, Zhao, Q, Kankanhalli, MS (2020-03-01). GradMix: Multi-source transfer across domains and tasks. Proceedings - 2020 IEEE Winter Conference on Applications of Computer Vision, WACV 2020 : 3008-3016. ScholarBank@NUS Repository. https://doi.org/10.1109/WACV45572.2020.9093343
Abstract: © 2020 IEEE. The computer vision community is witnessing an unprecedented rate of new tasks being proposed and addressed, thanks to the deep convolutional networks' capability to find complex mappings from X to Y. The advent of each task often accompanies the release of a large-scale annotated dataset, for supervised training of deep network. However, it is expensive and time-consuming to manually label sufficient amount of training data. Therefore, it is important to develop algorithms that can leverage off-the-shelf labeled dataset to learn useful knowledge for the target task. While previous works mostly focus on transfer learning from a single source, we study multi-source transfer across domains and tasks (MS-DTT), in a semi-supervised setting. We propose GradMix, a model-agnostic method applicable to any model trained with gradient-based learning rule, to transfer knowledge via gradient descent by weighting and mixing the gradients from all sources during training. GradMix follows a meta-learning objective, which assigns layer-wise weights to the source gradients, such that the combined gradient follows the direction that minimize the loss for a small set of samples from the target dataset. In addition, we propose to adaptively adjust the learning rate for each mini-batch based on its importance to the target task, and a pseudo-labeling method to leverage the unlabeled samples in the target domain. We conduct MS-DTT experiments on two tasks: digit recognition and action recognition, and demonstrate the advantageous performance of the proposed method against multiple baselines.
Source Title: Proceedings - 2020 IEEE Winter Conference on Applications of Computer Vision, WACV 2020
URI: https://scholarbank.nus.edu.sg/handle/10635/172483
DOI: 10.1109/WACV45572.2020.9093343
Appears in Collections:Staff Publications
Elements

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
2002.03264v1.pdf1.83 MBAdobe PDF

OPEN

Post-printView/Download

Google ScholarTM

Check

Altmetric


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.