Self-Balanced Dropout
arXiv:1908.01968
The paper demonstrates that conventional dropout does not fully eliminate co-adaptation due to input correlations and introduces Self-Balanced Dropout, a trainable dropout variant that mitigates this issue and improves performance across various tasks.
Abstract
Dropout is known as an effective way to reduce overfitting via preventing co-adaptations of units. In this paper, we theoretically prove that the co-adaptation problem still exists after using dropout due to the correlations among the inputs. Based on the proof, we further propose Self-Balanced Dropout, a novel dropout method which uses a trainable variable to balance the influence of the input correlation on parameter update. We evaluate Self-Balanced Dropout on a range of tasks with both simple and complex models. The experimental results show that the mechanism can effectively solve the co-adaption problem to some extent and significantly improve the performance on all tasks.