Jeremy Lin
Convnet is a GPU-based convolutional neural network Open Source code (C++11), written by the Geoffrey Hinton Deep learning team at the University of Toronto, whose original version was Hinton's student Alex Krizhevsky written by Cuda-convnet (whose project address is inGoogle Codeabove), the recent cuda-convnet has also been updated from version 1.0 to version 2.0 (Address).
The official address of this open source code is: Http://deeplearning.cs.toronto.edu/codes
The most famous in the open source code of CNN are two, one is Berkeley Caffe, the other is Toronto Convnet. Berkeley of the Caffe I have not studied its code, and has not been specifically used, bad comment. As for Toronto's convnet, I spent a lot of time in the first two weeks to see the source code of Cuda-convnet, in general, really painful to see, personally feel the structure of the code is a bit complex, to want to fully thoroughly understand is a need to spend some effort. Recently Toronto released a refactoring of the ConvNet1.0 source code, I looked at a cursory, found that the code is much clearer than the cuda-convnet, so I intend to be in the next period of time, to tidy up some of the ConvNet1.0 code reading notes.
here is a result of ConvNet1.0:
This address: http://blog.csdn.net/linj_m/article/details/38072145
More resources please follow blog:linjm-Machine Vision Weibo: Lin Jianmin-Machine Vision
