Class sparsity signature based Restricted Boltzmann Machine

作者：

Highlights：

• A supervised and semi-supervised approach to the RBM using class sparsity signature.

• Results of cssDBM and cssDBN along with dropout and dropconnect on MNIST, and CIFAR-10 databases.

• State-of-the-art results on one of the most challenging face database, Point and Shoot Challenge database.

摘要

Highlights•A supervised and semi-supervised approach to the RBM using class sparsity signature.•Results of cssDBM and cssDBN along with dropout and dropconnect on MNIST, and CIFAR-10 databases.•State-of-the-art results on one of the most challenging face database, Point and Shoot Challenge database.

论文关键词：Deep learning,Sparsity,Regularization,Object recognition

论文评审过程：Received 4 January 2016, Revised 27 April 2016, Accepted 29 April 2016, Available online 12 May 2016, Version of Record 13 October 2016.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.04.014