[docs] add note on Caffe convolution

diff --git a/docs/tutorial/convolution.md b/docs/tutorial/convolution.md
index a8bd0e6..2119d64 100644 (file)
--- a/docs/tutorial/convolution.md
+++ b/docs/tutorial/convolution.md
@@ -2,3 +2,12 @@
 layout: default
 ---
 # Caffeinated Convolution
+
+The Caffe strategy for convolution is to reduce the problem to matrix-matrix multiplication.
+This linear algebra computation is highly-tuned in BLAS libraries and efficiently computed on GPU devices.
+
+For more details read Yangqing's [Convolution in Caffe: a memo](https://github.com/Yangqing/caffe/wiki/Convolution-in-Caffe:-a-memo).
+
+As it turns out, this same reduction was independently explored in the context of conv. nets by
+
+> K. Chellapilla, S. Puri, P. Simard, et al. High performance convolutional neural networks for document processing. In Tenth International Workshop on Frontiers in Handwriting Recognition, 2006.

diff --git a/docs/tutorial/index.md b/docs/tutorial/index.md
index e9f593e..5638e41 100644 (file)
--- a/docs/tutorial/index.md
+++ b/docs/tutorial/index.md
@@ -28,6 +28,10 @@ and these principles direct the project.
 - [Layer Catalogue](layers.html): the layer is the fundamental unit of modeling and computation -- Caffe's catalogue includes layers for state-of-the-art models.
 - [Interfaces](interfaces.html): command line, Python, and MATLAB Caffe.
 
+For a closer look at a few details:
+
+- [Caffeinated Convolution](convolution.html): how Caffe computes convolutions.
+
 ## Deeper Learning
 
 There are helpful references freely online for deep learning that complement our hands-on tutorial.