Regularizing Meta-learning via Gradient Dropout

Hung Yu Tseng*, Yi Wen Chen, Yi Hsuan Tsai, Sifei Liu, Yen Yu Lin, Ming Hsuan Yang


研究成果: Conference contribution同行評審


With the growing attention on learning-to-learn new tasks using only a few examples, meta-learning has been widely used in numerous problems such as few-shot classification, reinforcement learning, and domain generalization. However, meta-learning models are prone to overfitting when there are no sufficient training tasks for the meta-learners to generalize. Although existing approaches such as Dropout are widely used to address the overfitting problem, these methods are typically designed for regularizing models of a single task in supervised training. In this paper, we introduce a simple yet effective method to alleviate the risk of overfitting for gradient-based meta-learning. Specifically, during the gradient-based adaptation stage, we randomly drop the gradient in the inner-loop optimization of each parameter in deep neural networks, such that the augmented gradients improve generalization to new tasks. We present a general form of the proposed gradient dropout regularization and show that this term can be sampled from either the Bernoulli or Gaussian distribution. To validate the proposed method, we conduct extensive experiments and analysis on numerous computer vision tasks, demonstrating that the gradient dropout regularization mitigates the overfitting problem and improves the performance upon various gradient-based meta-learning frameworks.

主出版物標題Computer Vision – ACCV 2020 - 15th Asian Conference on Computer Vision, 2020, Revised Selected Papers
編輯Hiroshi Ishikawa, Cheng-Lin Liu, Tomas Pajdla, Jianbo Shi
發行者Springer Science and Business Media Deutschland GmbH
出版狀態Published - 2021
事件15th Asian Conference on Computer Vision, ACCV 2020 - Virtual, Online
持續時間: 30 十一月 20204 十二月 2020


名字Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
12625 LNCS


Conference15th Asian Conference on Computer Vision, ACCV 2020
城市Virtual, Online


深入研究「Regularizing Meta-learning via Gradient Dropout」主題。共同形成了獨特的指紋。