ubuntu+python2。7下配置scikit-neuralnetwork

如何在ubuntu系统下配置神经网络学习包scikit-neuralnetwork

需要安装以下包

numpy，scipy，theano,pylearn2

1:安装numpy，scipy，theano

安装pip工具：sudo apt install python-pip

安装以下包：sudo apt install python2.7-dev

                        sudo apt upgrade gcc

                        sudo apt update

                        sudo pip install nose

                        sudo apt install git

安装numpy:sudo pip install numpy

测试:python -c "import numpy;numpy.test()"

安装scipy:sudo pip install scipy

测试：python - c "import scipy;scipy.test()"

安装Theano：sudo pip install Theano

测试：python -c "import theano;theano.test()"

提示没有parameterized

解决 sudo pip install nose-parameterized

2:pylearn2安装

下载：sudo git clone git://github.com/lisa-lab/pylearn2.git

安装：进到下载目录 cd pylearn2

           sudo python setup.py develop

3:安装scikit-neuralnetwork

sudo pip install matplotlib

sudo pip install scikit-neuralnetwork

安装完成

测试：nosetests -v sknn.tests

提示can't import name nonlinearities

解决sudo pip install --upgrade https://github.com/Lasagne/archive/master.zip

Lasagne版本过高，安装lasagne0.2.dev1

提示没有pandas

解决 sudo pip install pandas

提示 a length-2 tuple must be passed in for shape

解决sudo pip install scipy==0.17.1

提示没有python-tk

解决 sudo apt install python-tk

测试python example/plot_mlp.py --params activation

测试python example/bench_mnist.py sknn

测试时要进入scikit-neuralnetwork目录

下载地址：

git clone https://github.com/aigamedev/scikit-neuralnetwork.git