An example of MNIST handwritten numeral recognition and classification by using LSTM in Python

Time:2021-3-21

The code is as follows. I think the most important thing for beginners is to learn the format of RNN reading data.

# -*- coding: utf-8 -*-
"""
Created on Tue Oct 9 08:53:25 2018
@author: www
"""
 
import sys
sys.path.append('..')
 
import torch
import datetime
from torch.autograd import Variable
from torch import nn
from torch.utils.data import DataLoader
 
from torchvision import transforms as tfs
from torchvision.datasets import MNIST
 
#Defining data
data_tf = tfs.Compose([
   tfs.ToTensor(),
   tfs.Normalize([0.5], [0.5])
])
train_set = MNIST('E:/data', train=True, transform=data_tf, download=True)
test_set = MNIST('E:/data', train=False, transform=data_tf, download=True)
 
train_data = DataLoader(train_set, 64, True, num_workers=4)
test_data = DataLoader(test_set, 128, False, num_workers=4)
 
#Define the model
class rnn_classify(nn.Module):
   def __init__(self, in_feature=28, hidden_feature=100, num_class=10, num_layers=2):
     super(rnn_classify, self).__init__()
     self.rnn  =  nn.LSTM (in_ feature, hidden_ feature, num_ Layers) # using two layers LSTM
     self.classifier  =  nn.Linear (hidden_ feature, num_ Class) # the last RNN is fully connected to the final output
     
   def forward(self, x):
     #The size of X is (batch, 1, 28, 28), so we need to convert it to RNN input format (28, batch, 28)
     X = X. squeeze () # remove 1 from (batch, 1,28,28) and change to (batch, 28,28)
     X = X. permute (2, 0, 1) # put the last dimension into the first dimension, and become (batch, 28, 28)
     out, _  =  self.rnn (x) Using the default hidden state, the out is (28, batch, hidden)_ feature)
     Out = out [- 1,:,:] # take the last one in the sequence, and the size is (batch, hidden)_ feature)
     out =  self.classifier (out) # get the classification result
     return out
     
net = rnn_classify()
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.Adadelta(net.parameters(), 1e-1)
 
#Define the training process
def get_acc(output, label):
  total = output.shape[0]
  _, pred_label = output.max(1)
  num_correct = (pred_label == label).sum().item()
  return num_correct / total
  
  
def train(net, train_data, valid_data, num_epochs, optimizer, criterion):
  if torch.cuda.is_available():
    net = net.cuda()
  prev_time = datetime.datetime.now()
  for epoch in range(num_epochs):
    train_loss = 0
    train_acc = 0
    net = net.train()
    for im, label in train_data:
      if torch.cuda.is_available():
        im = Variable(im.cuda()) # (bs, 3, h, w)
        label = Variable(label.cuda()) # (bs, h, w)
      else:
        im = Variable(im)
        label = Variable(label)
      # forward
      output = net(im)
      loss = criterion(output, label)
      # backward
      optimizer.zero_grad()
      loss.backward()
      optimizer.step()
 
      train_loss += loss.item()
      train_acc += get_acc(output, label)
 
    cur_time = datetime.datetime.now()
    h, remainder = divmod((cur_time - prev_time).seconds, 3600)
    m, s = divmod(remainder, 60)
    time_str = "Time %02d:%02d:%02d" % (h, m, s)
    if valid_data is not None:
      valid_loss = 0
      valid_acc = 0
      net = net.eval()
      for im, label in valid_data:
        if torch.cuda.is_available():
          im = Variable(im.cuda())
          label = Variable(label.cuda())
        else:
          im = Variable(im)
          label = Variable(label)
        output = net(im)
        loss = criterion(output, label)
        valid_loss += loss.item()
        valid_acc += get_acc(output, label)
      epoch_str = (
        "Epoch %d. Train Loss: %f, Train Acc: %f, Valid Loss: %f, Valid Acc: %f, "
        % (epoch, train_loss / len(train_data),
          train_acc / len(train_data), valid_loss / len(valid_data),
          valid_acc / len(valid_data)))
    else:
      epoch_str = ("Epoch %d. Train Loss: %f, Train Acc: %f, " %
             (epoch, train_loss / len(train_data),
             train_acc / len(train_data)))
    prev_time = cur_time
    print(epoch_str + time_str)
    
train(net, train_data, test_data, 10, optimizer, criterion)

The above example of using LSTM to do MNIST handwritten numeral recognition classification by Python is the whole content shared by Xiaobian. I hope it can give you a reference and support developer.