pytorch入门级教程——分类代码分析与实现（iris数据集）

用iris数据进行分类训练，并可视化

首先导入API：

import torch
import torch.nn.functional as F
import matplotlib.pyplot as plt
from sklearn.decomposition import PCA
from torch.autograd import Variable
from sklearn.datasets import load_iris
import pandas as pd
import numpy as np

获得数据集：

iris = load_iris()
iris_d = pd.DataFrame(iris['data'], columns = ['Sepal_Length', 'Sepal_Width', 'Petal_Length', 'Petal_Width'])
iris_d['Species'] = iris.target

特征降维——主成分分析（PCA），将四维降至两维

transfer_1 = PCA(n_components=2)
iris_d = transfer_1.fit_transform(iris_d)
x = torch.from_numpy(iris_d)
y =torch.from_numpy(iris.target)
x, y = Variable(x), Variable(y)

构建前向神经网络，2个输入神经元，10个中间层神经元，3个输出神经元，激活函数使用RELU：

net =torch.nn.Sequential(
    torch.nn.Linear(2, 10),
    torch.nn.ReLU(),
    torch.nn.Linear(10, 3),
    )
print(net) 

设定autograd为梯度下降法以及损失函数为交叉熵误差：

optimizer = torch.optim.SGD(net.parameters(), lr=0.2)  #随机梯度下降
loss_func = torch.nn.CrossEntropyLoss() 

训练模型，循环150次：

for t in range(150):
    out = net(x.float())                 # input x and predict based on x
    loss = loss_func(out, y.long())     # must be (1. nn output, 2. target), the target label is NOT one-hotted

    optimizer.zero_grad()   # clear gradients for next train
    loss.backward()         # backpropagation, compute gradients
    optimizer.step()        # apply gradients

可视化，每十步更新一次：

if t % 10 == 0:
        # plot and show learning process
        plt.cla()
        prediction = torch.max(out, 1)[1]
        pred_y = prediction.data.numpy()
        target_y = y.data.numpy()
        plt.scatter(x.data.numpy()[:, 0], x.data.numpy()[:, 1], c=pred_y, s=100, lw=0, cmap='RdYlGn')
        accuracy = float((pred_y == target_y).astype(int).sum()) / float(target_y.size)
        plt.text(1.5, -4, 'Accuracy=%.2f' % accuracy, fontdict={'size': 20, 'color':  'red'})
        plt.pause(0.1)
plt.show()

保存及加载神经网络模型：

torch.save(net, 'E:\\net.pkl')  # save entire net
torch.save(net.state_dict(), 'E:\\net_params.pkl')   # save only the parameters
net1 = torch.load('net.pkl')  #加载神经网络

附完整代码：

import torch
import torch.nn.functional as F
import matplotlib.pyplot as plt
from sklearn.decomposition import PCA
from torch.autograd import Variable

# make fake data
from sklearn.datasets import load_iris
import pandas as pd
import numpy as np
iris = load_iris()
iris_d = pd.DataFrame(iris['data'], columns = ['Sepal_Length', 'Sepal_Width', 'Petal_Length', 'Petal_Width'])
iris_d['Species'] = iris.target
 
#特征降维——主成分分析
transfer_1 = PCA(n_components=2)
iris_d = transfer_1.fit_transform(iris_d)
x = torch.from_numpy(iris_d)
y =torch.from_numpy(iris.target)
x, y = Variable(x), Variable(y)

# mehod 
net =torch.nn.Sequential(
    torch.nn.Linear(2, 10),
    torch.nn.ReLU(),
    torch.nn.Linear(10, 3),
    )
#net1 = Net(n_feature=2, n_hidden=10, n_output=3)     # define the network
print(net)  # net architecture

optimizer = torch.optim.SGD(net.parameters(), lr=0.2)  #随机梯度下降
loss_func = torch.nn.CrossEntropyLoss()  # the target label is NOT an one-hotted



for t in range(150):
    out = net(x.float())                 # input x and predict based on x
    loss = loss_func(out, y.long())     # must be (1. nn output, 2. target), the target label is NOT one-hotted

    optimizer.zero_grad()   # clear gradients for next train
    loss.backward()         # backpropagation, compute gradients
    optimizer.step()        # apply gradients

    if t % 10 == 0:
        # plot and show learning process
        plt.cla()
        prediction = torch.max(out, 1)[1]
        pred_y = prediction.data.numpy()
        target_y = y.data.numpy()
        plt.scatter(x.data.numpy()[:, 0], x.data.numpy()[:, 1], c=pred_y, s=100, lw=0, cmap='RdYlGn')
        accuracy = float((pred_y == target_y).astype(int).sum()) / float(target_y.size)
        plt.text(1.5, -4, 'Accuracy=%.2f' % accuracy, fontdict={'size': 20, 'color':  'red'})
        plt.pause(0.1)


plt.show()
torch.save(net, 'E:\\net.pkl')  # save entire net
torch.save(net.state_dict(), 'E:\\net_params.pkl')   # save only the parameters
net1 = torch.load('net.pkl')  #加载神经网络

训练结果展示：