Partitioned Neural Networks Conference Paper uri icon

abstract

  • A new method is given for speeding up learning in a deep neural network with many hidden layers, by partially partitioning the network rather than fully interconnecting the layers. Empirical results are shown both for learning a simple Boolean function on a standard back-prop network, and for learning two different, complex, real-world vision tasks on a more sophisticated convolutional network. In all cases, the performance of the proposed system was better than traditional systems. The partially-partitioned network outperformed both the fully-partitioned and fully-unpartitioned networks.

author list (cited authors)

  • Sutton, D. P., Carlisle, M. C., Sarmiento, T. A., & Baird, L. C.

citation count

  • 0

publication date

  • June 2009

publisher