Processors White Papers
High Performance Convolutional Neural Networks for Document Processing
Overview Convolutional Neural Networks (CNNs) are well known for producing state-of-the-art recognizers for document processing. However, they can be difficult to implement and are usually slower than traditional Multi-Layer Perceptrons (MLPs). This paper presents three novel approaches to speeding up CNNs: unrolling convolution, using BLAS (basic linear algebra subroutines), and using GPUs (Graphic Processing Units). Unrolled convolution converts the processing in each convolutional layer (both forward-propagation and back-propagation) into a matrix-matrix product. The matrix-matrix product representation of CNNs makes their implementation as easy as MLPs. BLAS is used to efficiently compute matrix products on the CPU.
| Publisher | Microsoft | File Format | |
|---|---|---|---|
| Date Published | February 2006 | Downloads | 1 |
| Format | White Papers | ||
| Topics | |||
-
Did you find this white paper useful?
White Paper RSS
Download
Register
