This is Part 2 of the PyTorch Primer Series.
Linear Regression is linear approach for modeling the relationship between inputs and the predictions
Source: Wikipedia
We find a ‘Linear fit’ to the data.
Fit: We are trying to predict a variable y, by fitting a curve (line here) to the data. The curve in linear regression follows a linear relationship between the scalar (x) and dependent variable.
Creating Models in PyTorch
- Create a Class
- Declare your Forward Pass
- Tune the HyperParameters
class LinearRegressionModel(nn.Module):
**def** \_\_init\_\_(self, input\_dim, output\_dim):
super(LinearRegressionModel, self).\_\_init\_\_()
_\# Calling Super Class's constructor_
self.linear = nn.Linear(input\_dim, output\_dim)
_\# nn.linear is defined in nn.Module_
**def** forward(self, x):
_\# Here the forward pass is simply a linear function_
out = self.linear(x)
**return** out
input_dim = 1output_dim = 1
Steps
- Create instance of model
- Select Loss Criterion
- Choose Hyper Parameters
model = LinearRegressionModel(input_dim,output_dim)
criterion = nn.MSELoss()# Mean Squared Lossl_rate = 0.01optimiser = torch.optim.SGD(model.parameters(), lr = l_rate) #Stochastic Gradient Descent
epochs = 2000
Training The Model
for epoch in range(epochs):
epoch +=1
#increase the number of epochs by 1 every time
inputs = Variable(torch.from\_numpy(x\_train))
labels = Variable(torch.from\_numpy(y\_correct))
_#clear grads as discussed in prev post_
optimiser.zero\_grad()
_#forward to get predicted values_
outputs = model.forward(inputs)
loss = criterion(outputs, labels)
loss.backward()_\# back props_
optimiser.step()_\# update the parameters_
print('epoch **{}**, loss **{}**'.format(epoch,loss.data\[0\]))
Finally, Print the Predicted Values
predicted =model.forward(Variable(torch.from_numpy(x_train))).data.numpy()
plt.plot(x_train, y_correct, 'go', label = 'from data', alpha = .5)plt.plot(x_train, predicted, label = 'prediction', alpha = 0.5)plt.legend()plt.show()print(model.state_dict())
If you want to read about Week 2 in my Self Driving Journey, here is the blog post
The Next Part in the Series will discuss about Linear Regression.
You can find me on Twitter @bhutanisanyam1, connect with me on Linkedin here
Subscribe to my Newsletter for a weekly curated list of Deep Learning and Computer Vision Reads