Towards adaptive learning with improved convergence of deep belief networks on graphics processing units

作者：

Highlights：

• Adaptive step size technique that enhances the convergence of RBMs and DBNs.

• GPU parallel implementation of the RBMs and DBNs.

• Extensive experiment involving training hundreds of DBNs (MNIST and HHreco datasets).

摘要

Highlights•Adaptive step size technique that enhances the convergence of RBMs and DBNs.•GPU parallel implementation of the RBMs and DBNs.•Extensive experiment involving training hundreds of DBNs (MNIST and HHreco datasets).

论文关键词：Deep learning,Deep belief networks,Restricted Boltzmann machines,Contrastive divergence,Adaptive step size,GPU computing

论文评审过程：Available online 3 July 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.06.029