Shulou(Shulou.com)05/31 Report--

This article introduces the relevant knowledge of "how to use PyTorch for regression operation in python". In the operation of actual cases, many people will encounter such a dilemma, so let the editor lead you to learn how to deal with these situations. I hope you can read it carefully and be able to achieve something!

Important basic functions in PyTorch 1. Construction of class Net (torch.nn.Module) neural network:

The construction of neural network in PyTorch is different from that of Tensorflow. It needs to be built with a class (later, it can be built with a Sequential model similar to Keras). Of course, the foundation is still built with classes, which need to inherit the neural network model in PyTorch, torch.nn.Module. The specific construction method is as follows:

# inherit torch.nn.Module model class Net (torch.nn.Module): # overload initialization function (I forgot whether this is called overload) def _ _ init__ (self, n_feature, n_hidden, n_output): super (Net, self). _ _ init__ () # Applies a linear transformation to the incoming data:: math:y = Xa ^ T + b # full connection layer The formula is y = Xa ^ T + b # to build two fully connected layers (that is, a hidden layer) self.hidden = torch.nn.Linear (n_feature, n_hidden) self.predict = torch.nn.Linear (n_hidden, n_output) # forward function to build the forward pass procedure def forward (self) while initializing X): # output of hidden layer hidden_layer = functional.relu (self.hidden (x)) # actual output output_layer = self.predict (hidden_layer) return output_layer

In this part, a neural network with a hidden layer is constructed, and the number of neurons in the hidden layer is n_hidden.

After the above classes are established, the neural network can be established with the following functions:

Net = Net (n_feature=1, n_hidden=10, n_output=1) 2, optimizer optimizer

The optimizer used by optimizer to build the model has the same meaning as the optimizer in tensorflow, which is in the library prefixed with torch.optim.

The optimizer needs to pass in parameters of the net network.

The specific usage is as follows:

# torch.optim is the optimizer module # Adam can be changed to other optimizers, such as SGD, RMSprop, etc. Optimizer = torch.optim.Adam (net.parameters (), lr=1e-3) 3, loss loss function definition

Loss is used to define the loss function of neural network training. The commonly used loss functions are mean square error loss function (regression) and cross-entropy loss function (classification).

The specific usage is as follows:

# mean square error lossloss_func = torch.nn.MSELoss () 4. Training process

The training process is divided into three steps:

1. Use the network to predict the results.

Prediction = net (x)

2. Loss is generated by comparing the predicted results with the real values.

Loss = loss_func (prediction, y)

3. Reverse transfer (this part has three steps).

# mean square error loss# reverse transfer step # 1, initialize gradient optimizer.zero_grad () # 2, calculate gradient loss.backward () # 3, optimize optimizer.step () for optimizer

This is a simple regression prediction model.

The shape of import torchfrom torch.autograd import Variableimport torch.nn.functional as functionalimport matplotlib.pyplot as pltimport numpy as np# x is (100jue 1) x = torch.from_numpy (np.linspace (- 1jue 1100). Reshape ([100jue 1]). The shape of type (torch.FloatTensor) # y is (100jue 1) y = torch.sin (x) + 0.2*torch.rand (x.size ()) class Net (torch.nn.Module): def _ init__ (self, n_feature, n_hidden) N_output): super (Net, self). _ _ init__ () # Applies a linear transformation to the incoming data:: math:y = Xa ^ T + b # full connection layer The formula is y = Xa ^ T + b self.hidden = torch.nn.Linear (n_feature, n_hidden) self.predict = torch.nn.Linear (n_hidden, n_output) def forward (self, x): # output of the hidden layer hidden_layer = functional.relu (self.hidden (x)) output_layer = self.predict (hidden_layer) return output_layer# Class Establishment net = Net (n_feature=1 N_hidden=10, n_output=1) # torch.optim is the optimizer module optimizer = torch.optim.Adam (net.parameters (), lr=1e-3) # mean square error lossloss_func = torch.nn.MSELoss () for t in range (1000): prediction = net (x) loss = loss_func (prediction) Y) # reverse transfer step # 1, initialize gradient optimizer.zero_grad () # 2, calculate gradient loss.backward () # 3, optimize optimizer optimizer.step () if t & 50 = = 0: print ("The loss is", loss.data.numpy ())

The running result is:

The loss is 0.27913737

The loss is 0.2773982

The loss is 0.27224126

.

The loss is 0.0035993527

The loss is 0.0035974088

The loss is 0.0035967692

This is the end of the content of "how to use PyTorch for regression operation in python". Thank you for your reading. If you want to know more about the industry, you can follow the website, the editor will output more high-quality practical articles for you!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.