Pytorch理解更多神經網絡優化方法

PyTorch種優化器選擇
SGD、Momentum、RMSprop、Adam

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

# 超參數
torch.manual_seed(1)
LR = 0.01
Batch_size = 24
Epoch = 24

# 定義數據集
x = torch.unsqueeze(torch.linspace(-1,1,1000),dim=1)
y = x**2 + 0.1*torch.normal(torch.zeros(x.size()))

# 先轉換成torch能識別的Dataset >> 再使用數據加載器加載數據
data_set = torch.utils.data.TensorDataset(x,y)
dataset_loader = torch.utils.data.DataLoader(dataset = data_set,
                                             batch_size = Batch_size,
                                             shuffle = True,
                                             num_workers = 2,)
# 定義神經網絡
class Net(torch.nn.Module):
    def __init__(self):
        super(Net,self).__init__()
        self.hidden = torch.nn.Linear(1,20)
        self.predict = torch.nn.Linear(20,1)

    def forward(self, input):
        x = F.relu(self.hidden(input))
        x = self.predict(x)
        return x

net_SGD = Net()
net_Momentum = Net()
net_RMSprop = Net()
net_Adam = Net()
nets = [net_SGD,net_Momentum,net_RMSprop,net_Adam]

# 定義優化器（學習率相同）
opt_SGD         = torch.optim.SGD(net_SGD.parameters(), lr=LR)
opt_Momentum    = torch.optim.SGD(net_Momentum.parameters(), lr=LR, momentum=0.8)
opt_RMSprop     = torch.optim.RMSprop(net_RMSprop.parameters(), lr=LR, alpha=0.9)
opt_Adam        = torch.optim.Adam(net_Adam.parameters(), lr=LR, betas=(0.9, 0.99))
optimizers = [opt_SGD, opt_Momentum, opt_RMSprop, opt_Adam] # 放在list裏面，可以用個for循環遍歷

#迴歸誤差
loss_func = torch.nn.MSELoss()
losses_his = [[], [], [], []]                  # 記錄 training 時不同神經網絡的 loss


if __name__ == '__main__':
    for epoch in range(Epoch):
        print('Epoch: ', epoch)
        for step, (batch_x, batch_y) in enumerate(dataset_loader):
            b_x = Variable(batch_x)            # 包裝成Variable
            b_y = Variable(batch_y)

            # 對每個優化器, 優化屬於他的神經網絡
            for net, opt, l_his in zip(nets, optimizers, losses_his):  # 三個都是list形式zip打包處理
                output = net(b_x)              # get output for every net
                loss = loss_func(output, b_y)  # compute loss for every net
                opt.zero_grad()                # clear gradients for next train
                loss.backward()                # backpropagation, compute gradients
                opt.step()                     # apply gradients
                l_his.append(loss.data.item()) # loss recoder

        labels = ['SGD', 'Momentum', 'RMSprop', 'Adam']
        for i, l_his in enumerate(losses_his):
            plt.plot(l_his, label=labels[i])
        plt.legend(loc='best')
        plt.xlabel('Steps')
        plt.ylabel('Loss')
        plt.ylim(0, 0.2)
        plt.show()