6、Multiple Dimension Input 处理多维特征的输入

B站视频教程传送门：PyTorch深度学习实践 - 处理多维特征的输入

6.1 Revision

我们先来回顾一下回归与分类：

差别： 主要在于输出值

回归（Regressiom）：y ∈ R

分类（Classification）：y ∈ { } 离散的集合

6.2 Diabetes Dataset 糖尿病数据集

如果我们安装过sklearn（python编程安装sklearn），其中就包含糖尿病数据集，可以进入该目录（D:\Software\Anaconda\Lib\site-packages\sklearn\datasets\data）下查看，如下图所示：

6.3 Logistic Regression Model 逻辑斯蒂回归模型

由于这里的 x不再是简简单单的一维，而是 8维，所以应该看成下方两个矩阵相乘：

注：$\sigma =\frac{1}{1+e^{-z}}$

6.4 Mini-Batch（N samples）

import torch


class Liang(torch.nn.Module):
    def __init__(self):
        super(Liang, self).__init__()
        self.linear = nn.Linear(8, 1)
        self.sigmoid = torch.nn.Sigmoid()

    def forward(self, x):
        x = self.sigmoid(self.linear(x))
        return x


model = Liang()

6.5 Neural Network 神经网络

当输入8维，输出2维时：

self.linear = torch.nn.Linear(8, 2)

当输入8维，输出6维时：

self.linear = torch.nn.Linear(8, 6)

可以降维，可以升维，也可以一降（升）一升（降）：

6.6 Diabetes Prediction 糖尿病预测

X1~X8：病人相应的指标

Y：一年后病情是否加重（预测）

6.6.1 Prepare Dataset

import numpy as np

xy = np.loadtxt('../data/diabetes.csv.gz', delimiter=',', dtype=np.float32)
x_data = torch.from_numpy(xy[:, :-1])  # 所有行，除了最后一列
y_data = torch.from_numpy(xy[:, [-1]])  # 所有行，最后一列 转为矩阵而不是向量

6.6.2 Define Model

6.6.3 Construct Loss and Optimizer

criterion = torch.nn.BCELoss(reduction='mean')
optimizer = torch.optim.SGD(model.parameters(), lr=0.1)

6.6.4 Training Cycle

for epoch in range(100):
    # Forward
    y_pred = model(x_data) # This program has not use Mini-Batch for training. We shall talk about DataLoader later.
    loss = criterion(y_pred, y_data)
    print(epoch, loss.item())

    # Backward
    optimizer.zero_grad()
    loss.backward()
    
    # Update
    optimizer.step()

6.6.5 Activate function

神经网络中激活函数的可视化：https://dashee87.github.io/deep%20learning/visualising-activation-functions-in-neural-networks/

PyTorch文档：https://pytorch.org/docs/stable/nn.html#non-linear-activations-weighted-sum-nonlinearity

6.6.5 完整代码

import torch
import numpy as np
import matplotlib.pyplot as plt

xy = np.loadtxt('../data/diabetes.csv.gz', delimiter=',', dtype=np.float32)

x_data = torch.from_numpy(xy[:, :-1])  # 所有行，除了最后一列
y_data = torch.from_numpy(xy[:, [-1]])  # 所有行，最后一列 转为矩阵而不是向量


class Liang(torch.nn.Module):
    def __init__(self):
        super(Liang, self).__init__()
        self.linear1 = torch.nn.Linear(8, 6)
        self.linear2 = torch.nn.Linear(6, 4)
        self.linear3 = torch.nn.Linear(4, 1)

        self.sigmoid = torch.nn.Sigmoid()  # Sigmoid
        self.tanh = torch.nn.Tanh()

    def forward(self, x):
        x = self.tanh(self.linear1(x))
        x = self.tanh(self.linear2(x))
        x = self.sigmoid(self.linear3(x))
        return x


model = Liang()

criterion = torch.nn.BCELoss(reduction='mean')
optimizer = torch.optim.SGD(model.parameters(), lr=0.1)

epoch_list = []
loss_list = []

for epoch in range(100):
    # Forward
    y_pred = model(x_data)
    loss = criterion(y_pred, y_data)
    print(epoch, loss.item())

    epoch_list.append(epoch)
    loss_list.append(loss.item())

    # Backward
    optimizer.zero_grad()
    loss.backward()

    # Update
    optimizer.step()

plt.plot(epoch_list, loss_list)
plt.xlabel('Epoch')
plt.ylabel('Loss')
plt.title('Tanh')
plt.show()