Py之neurolab：Python库之neurolab的简介、安装、使用方法之详细攻略

neurolab的简介

   neurolab是一个简单而强大的Python神经网络库。包含基于神经网络、训练算法和灵活的框架来创建和探索其他神经网络类型。NeuroLab一个具有灵活网络配置和Python学习算法的基本神经网络算法库。为了简化库的使用，接口类似于MATLAB（C）的神经网络工具箱（NNT）的包。该库基于包NUMPY（http://NoPy.SimP.org），使用一些学习算法。

neurolab

neurolab的安装

pip install neurolab

neurolab的使用方法

Support neural networks types

Single layer perceptron

create function: neurolab.net.newp()

example of use: newp

default train function: neurolab.train.train_delta()

support train functions: train_gd, train_gda, train_gdm, train_gdx, train_rprop, train_bfgs, train_cg

Multilayer feed forward perceptron

create function: neurolab.net.newff()

example of use: newff

default train function: neurolab.train.train_gdx()

support train functions: train_gd, train_gda, train_gdm, train_rprop, train_bfgs, train_cg

Competing layer (Kohonen Layer)

create function: neurolab.net.newc()

example of use: newc

default train function: neurolab.train.train_cwta()

support train functions: train_wta

Learning Vector Quantization (LVQ)

create function: neurolab.net.newlvq()

example of use: newlvq

default train function: neurolab.train.train_lvq()

Elman Recurrent network

create function: neurolab.net.newelm()

example of use: newelm

default train function: neurolab.train.train_gdx()

support train functions: train_gd, train_gda, train_gdm, train_rprop, train_bfgs, train_cg

Hopfield Recurrent network

create function: neurolab.net.newhop()

example of use: newhop

Hemming Recurrent network

create function: neurolab.net.newhem()

example of use: newhem

Features:

Pure python + numpy

API like Neural Network Toolbox (NNT) from MATLAB

Interface to use train algorithms form scipy.optimize

Flexible network configurations and learning algorithms. You may change: train, error, initialization and activation functions

Unlimited number of neural layers and number of neurons in layers

Variety of supported types of Artificial Neural Network and learning algorithms

Example:

>>> import numpy as np

>>> import neurolab as nl

>>> # Create train samples

>>> input = np.random.uniform(-0.5, 0.5, (10, 2))

>>> target = (input[:, 0] + input[:, 1]).reshape(10, 1)

>>> # Create network with 2 inputs, 5 neurons in input layer and 1 in output layer

>>> net = nl.net.newff([[-0.5, 0.5], [-0.5, 0.5]], [5, 1])

>>> # Train process

>>> err = net.train(input, target, show=15)

Epoch: 15; Error: 0.150308402918;

Epoch: 30; Error: 0.072265865089;

Epoch: 45; Error: 0.016931355131;

The goal of learning is reached

>>> # Test

>>> net.sim([[0.2, 0.1]]) # 0.2 + 0.1

array([[ 0.28757596]])