LGAug 19, 2015
Dither is Better than Dropout for Regularising Deep Neural Networks
arXiv:1508.04826v212 citations
AI Analysis
This addresses the need for better regularization methods in deep learning, but appears incremental as it builds on existing techniques.
The paper tackled the problem of regularizing deep neural networks by comparing dropout with dither, showing that dither is more effective due to dropout's inherent limitations.
Regularisation of deep neural networks (DNN) during training is critical to performance. By far the most popular method is known as dropout. Here, cast through the prism of signal processing theory, we compare and contrast the regularisation effects of dropout with those of dither. We illustrate some serious inherent limitations of dropout and demonstrate that dither provides a more effective regulariser.