(转:http://blog.csdn.net/lucktroy/article/details/9466369


Pylearn2 是建立在Theano之上的一个机器学习库,可以参考:http://deeplearning.net/software/pylearn2/

早前我已经用《Install Theano in Linux without root authority (sodu)》,所以现在也只需要仿照着安装即可。

1. Download:

  1. $ git clone git://github.com/lisa-lab/pylearn2.git  

2. 配置环境变量,我一般用.bash_profile, 也可以放在.bashrc里

  1. export PYLEARN2_DATA_PATH=/path/to/data  

3. 安装,进入pylearn2

  1. $ python setup.py build  
  1. $ python setup.py install --prefix=$PYTHON_HOME  

4. 测试,可以看http://deeplearning.net/software/pylearn2/tutorial/index.html#tutorial,首先下载数据集cifar-10: http://www.cs.toronto.edu/~kriz/cifar.html

  1. $ wget http://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz  
  1. $ tar -zxvf cifar-10*  
4.1. 生成数据子集,进入 $PYLEARN2_HOME/pylearn2/scripts/tutorials/grbm_smd/

  1. $ python make_dataset.py  
4.2 训练模型

  1. $ python ../../train.py cifar_grbm_smd.yaml  
4.3 视察模型

  1. $ python ../../show_weights.py cifar_grbm_smd.pkl  


--------------------------------------------------------------------------------------------------

pylearn2官方链接:http://deeplearning.net/software/pylearn2/index.html


基于pylearn2的几个实验:(MNIST数据集http://yann.lecun.com/exdb/mnist/

http://nbviewer.ipython.org/urls/github.com/lisa-lab/pylearn2/raw/master/pylearn2/scripts/tutorials/softmax_regression.ipynb

http://nbviewer.ipython.org/urls/github.com/lisa-lab/pylearn2/raw/master/pylearn2/scripts/tutorials/multilayer_perceptron.ipynb

http://nbviewer.ipython.org/urls/raw.github.com/lisa-lab/pylearn2/master/pylearn2/scripts/tutorials/convolutional_network.ipynb

http://nbviewer.ipython.org/urls/raw.github.com/lisa-lab/pylearn2/master/pylearn2/scripts/tutorials/stacked_autoencoders.ipynb


pylearn2的更多细节介绍:http://arxiv.org/pdf/1308.4214v1.pdf



Logo

更多推荐