# Logistic Regression
# Q1 why not maximize accuracy?
如果最小化 accuracy,会造成 gradient = 0 的现象。因为用 sigmoid 做激活函数,小于 0.5 判断为 0,大于 0.5 判断为 1,如果参数 w 从 0.4 提升到 0.45,但因为还是小于 0.5,所以预测的类别没有发生变化,accuracy 也没有变,导致梯度不就为 0 嘛。
会造成梯度不连续的现象。还是刚才的例子,如果 w 从 0.499 变到 0.501,只有微小的提升,但是预测的类别就发生了变化,导致 accuracy 突变。也就造成了刚刚梯度都还是 0,但这次就突然变了很大的一个值,就是这种梯度不连续的现象。
所以在分类问题上,不会用 accuracy 作为优化目标。
# Q2 why call logistic regression
因为用了 sigmoid 激活函数,所以称它为 logistic
如果使用 MSE 损失函数,则可以称之为 regression,因为 MSE 就是想将比如 P (0.7) 逼近 P (1),所以可以称之为 regression
而如果用 CrossEntropy 作为损失函数,则应该称为 classification
# Cross Entropy
# Entropy
# CrossEntropy
p 定义为真实分布,q 是模型学习到的分布,则 CrossEntropy 的定义如下
\begin{align*}H\left(p,q\right)&=-\sum p\left(x\right)\log q\left(x\right)\\[10 pt] &=H\left(p\right)+D_{KL}\left(p\left|q\right.\right)\end{align*}KL 散度,用来衡量两个分布间的差异,详情:Kullback-Leibler (KL) 散度介绍
当 q = p,即完全学习到了真实分布时,CrossEntropy = Entropy
对于分类问题,通常使用 one-hot 编码,这种情况下,真实分布 p 就类似 [1, 0, 0, 0] 这种,H (p) 的值就是 entropy = 1log1 = 0,所以对于分类问题,CrossEntropy 就等于用 KL 散度来衡量分布间的差异
通常情况下,将没有经过激活函数的输出称为 Logits,如下
# 案例
注意,在 pytorch 中,cross_entropy 就等于 softmax + log + nll_loss,所以对于 logits,直接扔到 cross_entropy 中就行
# 全连接网络多分类小实战
# 底层 API 版
import torch | |
import torch.nn as nn | |
import torch.nn.functional as F | |
import torch.optim as optim | |
from torchvision import datasets, transforms | |
batch_size=200 | |
learning_rate=0.01 | |
epochs=10 | |
train_loader = torch.utils.data.DataLoader( | |
datasets.MNIST('../data', train=True, download=True, | |
transform=transforms.Compose([ | |
transforms.ToTensor(), | |
transforms.Normalize((0.1307,), (0.3081,)) | |
])), | |
batch_size=batch_size, shuffle=True) | |
test_loader = torch.utils.data.DataLoader( | |
datasets.MNIST('../data', train=False, transform=transforms.Compose([ | |
transforms.ToTensor(), | |
transforms.Normalize((0.1307,), (0.3081,)) | |
])), | |
batch_size=batch_size, shuffle=True) | |
w1, b1 = torch.randn(200, 784, requires_grad=True),\ | |
torch.zeros(200, requires_grad=True) | |
w2, b2 = torch.randn(200, 200, requires_grad=True),\ | |
torch.zeros(200, requires_grad=True) | |
w3, b3 = torch.randn(10, 200, requires_grad=True),\ | |
torch.zeros(10, requires_grad=True) | |
torch.nn.init.kaiming_normal_(w1) | |
torch.nn.init.kaiming_normal_(w2) | |
torch.nn.init.kaiming_normal_(w3) | |
def forward(x): | |
x = x@w1.t() + b1 | |
x = F.relu(x) | |
x = x@w2.t() + b2 | |
x = F.relu(x) | |
x = x@w3.t() + b3 | |
x = F.relu(x) | |
return x | |
optimizer = optim.SGD([w1, b1, w2, b2, w3, b3], lr=learning_rate) | |
criteon = nn.CrossEntropyLoss() | |
for epoch in range(epochs): | |
for batch_idx, (data, target) in enumerate(train_loader): | |
data = data.view(-1, 28*28) | |
logits = forward(data) | |
loss = criteon(logits, target) | |
optimizer.zero_grad() | |
loss.backward() | |
# print(w1.grad.norm(), w2.grad.norm()) | |
optimizer.step() | |
if batch_idx % 100 == 0: | |
print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format( | |
epoch, batch_idx * len(data), len(train_loader.dataset), | |
100. * batch_idx / len(train_loader), loss.item())) | |
test_loss = 0 | |
correct = 0 | |
for data, target in test_loader: | |
data = data.view(-1, 28 * 28) | |
logits = forward(data) | |
test_loss += criteon(logits, target).item() | |
pred = logits.data.max(1)[1] | |
correct += pred.eq(target.data).sum() | |
test_loss /= len(test_loader.dataset) | |
print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format( | |
test_loss, correct, len(test_loader.dataset), | |
100. * correct / len(test_loader.dataset))) |
# 高阶 API 版
# nn.ReLU 和 F.relu () 的区别
前者是类风格的 API,后者是函数风格的 API
类风格的要创建实例,然后扔进去。函数风格的可以直接调用
import torch | |
import torch.nn as nn | |
import torch.nn.functional as F | |
import torch.optim as optim | |
from torchvision import datasets, transforms | |
batch_size=200 | |
learning_rate=0.01 | |
epochs=10 | |
train_loader = torch.utils.data.DataLoader( | |
datasets.MNIST('../data', train=True, download=True, | |
transform=transforms.Compose([ | |
transforms.ToTensor(), | |
transforms.Normalize((0.1307,), (0.3081,)) | |
])), | |
batch_size=batch_size, shuffle=True) | |
test_loader = torch.utils.data.DataLoader( | |
datasets.MNIST('../data', train=False, transform=transforms.Compose([ | |
transforms.ToTensor(), | |
transforms.Normalize((0.1307,), (0.3081,)) | |
])), | |
batch_size=batch_size, shuffle=True) | |
class MLP(nn.Module): | |
def __init__(self): | |
super(MLP, self).__init__() | |
self.model = nn.Sequential( | |
nn.Linear(784, 200), | |
nn.ReLU(inplace=True), | |
nn.Linear(200, 200), | |
nn.ReLU(inplace=True), | |
nn.Linear(200, 10), | |
nn.ReLU(inplace=True), | |
) | |
def forward(self, x): | |
x = self.model(x) | |
return x | |
net = MLP() | |
optimizer = optim.SGD(net.parameters(), lr=learning_rate) | |
criteon = nn.CrossEntropyLoss() | |
for epoch in range(epochs): | |
for batch_idx, (data, target) in enumerate(train_loader): | |
data = data.view(-1, 28*28) | |
logits = net(data) | |
loss = criteon(logits, target) | |
optimizer.zero_grad() | |
loss.backward() | |
# print(w1.grad.norm(), w2.grad.norm()) | |
optimizer.step() | |
if batch_idx % 100 == 0: | |
print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format( | |
epoch, batch_idx * len(data), len(train_loader.dataset), | |
100. * batch_idx / len(train_loader), loss.item())) | |
test_loss = 0 | |
correct = 0 | |
for data, target in test_loader: | |
data = data.view(-1, 28 * 28) | |
logits = net(data) | |
test_loss += criteon(logits, target).item() | |
pred = logits.data.max(1)[1] | |
correct += pred.eq(target.data).sum() | |
test_loss /= len(test_loader.dataset) | |
print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format( | |
test_loss, correct, len(test_loader.dataset), | |
100. * correct / len(test_loader.dataset))) |
# 激活函数补充
# Leaky ReLU
# SELU
# Softplus
# GPU 加速
使用 cuda
会造成硬编码,所以推荐使用 to(device)
# MNIST 实战
import torch | |
import torch.nn as nn | |
import torch.nn.functional as F | |
import torch.optim as optim | |
from torchvision import datasets, transforms | |
batch_size=200 | |
learning_rate=0.01 | |
epochs=10 | |
train_loader = torch.utils.data.DataLoader( | |
datasets.MNIST('../data', train=True, download=True, | |
transform=transforms.Compose([ | |
transforms.ToTensor(), | |
transforms.Normalize((0.1307,), (0.3081,)) | |
])), | |
batch_size=batch_size, shuffle=True) | |
test_loader = torch.utils.data.DataLoader( | |
datasets.MNIST('../data', train=False, transform=transforms.Compose([ | |
transforms.ToTensor(), | |
transforms.Normalize((0.1307,), (0.3081,)) | |
])), | |
batch_size=batch_size, shuffle=True) | |
class MLP(nn.Module): | |
def __init__(self): | |
super(MLP, self).__init__() | |
self.model = nn.Sequential( | |
nn.Linear(784, 200), | |
nn.LeakyReLU(inplace=True), | |
nn.Linear(200, 200), | |
nn.LeakyReLU(inplace=True), | |
nn.Linear(200, 10), | |
nn.LeakyReLU(inplace=True), | |
) | |
def forward(self, x): | |
x = self.model(x) | |
return x | |
device = torch.device('cuda:0') | |
net = MLP().to(device) | |
optimizer = optim.SGD(net.parameters(), lr=learning_rate) | |
criteon = nn.CrossEntropyLoss().to(device) | |
for epoch in range(epochs): | |
for batch_idx, (data, target) in enumerate(train_loader): | |
data = data.view(-1, 28*28) | |
data, target = data.to(device), target.cuda() | |
logits = net(data) | |
loss = criteon(logits, target) | |
optimizer.zero_grad() | |
loss.backward() | |
# print(w1.grad.norm(), w2.grad.norm()) | |
optimizer.step() | |
if batch_idx % 100 == 0: | |
print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format( | |
epoch, batch_idx * len(data), len(train_loader.dataset), | |
100. * batch_idx / len(train_loader), loss.item())) | |
test_loss = 0 | |
correct = 0 | |
for data, target in test_loader: | |
data = data.view(-1, 28 * 28) | |
data, target = data.to(device), target.cuda() | |
logits = net(data) | |
test_loss += criteon(logits, target).item() | |
pred = logits.argmax(dim=1) | |
correct += pred.eq(target).float().sum().item() | |
test_loss /= len(test_loader.dataset) | |
print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format( | |
test_loss, correct, len(test_loader.dataset), | |
100. * correct / len(test_loader.dataset))) |
# Visdom 可视化
# 安装及运行
安装 pip install visdom
启动服务 python -m visdom.server
# 绘制单条曲线
# 绘制多条曲线
# 绘制图片和文本
import torch | |
import torch.nn as nn | |
import torch.nn.functional as F | |
import torch.optim as optim | |
from torchvision import datasets, transforms | |
from visdom import Visdom | |
batch_size=200 | |
learning_rate=0.01 | |
epochs=10 | |
train_loader = torch.utils.data.DataLoader( | |
datasets.MNIST('../data', train=True, download=True, | |
transform=transforms.Compose([ | |
transforms.ToTensor(), | |
# transforms.Normalize((0.1307,), (0.3081,)) | |
])), | |
batch_size=batch_size, shuffle=True) | |
test_loader = torch.utils.data.DataLoader( | |
datasets.MNIST('../data', train=False, transform=transforms.Compose([ | |
transforms.ToTensor(), | |
# transforms.Normalize((0.1307,), (0.3081,)) | |
])), | |
batch_size=batch_size, shuffle=True) | |
class MLP(nn.Module): | |
def __init__(self): | |
super(MLP, self).__init__() | |
self.model = nn.Sequential( | |
nn.Linear(784, 200), | |
nn.LeakyReLU(inplace=True), | |
nn.Linear(200, 200), | |
nn.LeakyReLU(inplace=True), | |
nn.Linear(200, 10), | |
nn.LeakyReLU(inplace=True), | |
) | |
def forward(self, x): | |
x = self.model(x) | |
return x | |
device = torch.device('cuda:0') | |
net = MLP().to(device) | |
optimizer = optim.SGD(net.parameters(), lr=learning_rate) | |
criteon = nn.CrossEntropyLoss().to(device) | |
viz = Visdom() | |
viz.line([0.], [0.], win='train_loss', opts=dict(title='train loss')) | |
viz.line([[0.0, 0.0]], [0.], win='test', opts=dict(title='test loss&acc.', | |
legend=['loss', 'acc.'])) | |
global_step = 0 | |
for epoch in range(epochs): | |
for batch_idx, (data, target) in enumerate(train_loader): | |
data = data.view(-1, 28*28) | |
data, target = data.to(device), target.cuda() | |
logits = net(data) | |
loss = criteon(logits, target) | |
optimizer.zero_grad() | |
loss.backward() | |
# print(w1.grad.norm(), w2.grad.norm()) | |
optimizer.step() | |
global_step += 1 | |
viz.line([loss.item()], [global_step], win='train_loss', update='append') | |
if batch_idx % 100 == 0: | |
print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format( | |
epoch, batch_idx * len(data), len(train_loader.dataset), | |
100. * batch_idx / len(train_loader), loss.item())) | |
test_loss = 0 | |
correct = 0 | |
for data, target in test_loader: | |
data = data.view(-1, 28 * 28) | |
data, target = data.to(device), target.cuda() | |
logits = net(data) | |
test_loss += criteon(logits, target).item() | |
pred = logits.argmax(dim=1) | |
correct += pred.eq(target).float().sum().item() | |
viz.line([[test_loss, correct / len(test_loader.dataset)]], | |
[global_step], win='test', update='append') | |
viz.images(data.view(-1, 1, 28, 28), win='x') | |
viz.text(str(pred.detach().cpu().numpy()), win='pred', | |
opts=dict(title='pred')) | |
test_loss /= len(test_loader.dataset) | |
print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format( | |
test_loss, correct, len(test_loader.dataset), | |
100. * correct / len(test_loader.dataset))) |