Lossless-constraint Denoising based Auto-encoders

作者：

Highlights：

•

• A Lossless-constraint Denoising (LD) method to enhance the anti-noise ability and robustness of auto-encoders is proposed.

• Respectively combine the Denoising Auto-encoder (DAE) and Sparse Auto-encoder (SAE) with LD method, design two auto-encoders of better noise immunity: Lossless-constraint Denoising Auto-encoder (LDAE) and Lossless-constraint Denoising Sparse Auto-encoder (LDSAE).

• Stack the LDSAE or LDAE to generate deep structure.

摘要

•A Lossless-constraint Denoising (LD) method to enhance the anti-noise ability and robustness of auto-encoders is proposed.•Respectively combine the Denoising Auto-encoder (DAE) and Sparse Auto-encoder (SAE) with LD method, design two auto-encoders of better noise immunity: Lossless-constraint Denoising Auto-encoder (LDAE) and Lossless-constraint Denoising Sparse Auto-encoder (LDSAE).•Stack the LDSAE or LDAE to generate deep structure.

论文关键词：Sparse auto-encoder,Denoising auto-encoder,Neural networks and deep learning

论文评审过程：Received 31 May 2017, Revised 2 February 2018, Accepted 2 February 2018, Available online 9 February 2018, Version of Record 16 February 2018.

论文官网地址：https://doi.org/10.1016/j.image.2018.02.002