A weight perturbation-based regularisation technique for convolutional neural networks and the application in medical imaging

作者：

Highlights：

• A regularisation model based on noise perturbation for convolutional neural networks.

• Accelerated convergence speed, circumvent over-fitting, and improved generalisation.

• Applying additive noise to earlier convolutional layers achieves better performance.

摘要

•A regularisation model based on noise perturbation for convolutional neural networks.•Accelerated convergence speed, circumvent over-fitting, and improved generalisation.•Applying additive noise to earlier convolutional layers achieves better performance.

论文关键词：Convolutional Neural Network,Regularisation,Generalisation,Weight perturbation

论文评审过程：Received 19 February 2019, Revised 8 January 2020, Accepted 8 January 2020, Available online 22 January 2020, Version of Record 19 February 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.113196