Einführung Deep Learning

Allgemeine Tips

GPU in Google Colab einschalten: "Laufzeit -> Laufzeittyp ändern -> Hardwarebeschleuniger -> T4 GPU"

Aufgabe

Trainiere einen eigenen Bildklassifikator mit zwei (oder mehreren Klassen) deiner Wahl (z.B. Hunde vs. Katzen, Hotdogs vs. Burger etc. :D). Nutze die fast.ai Bibliothek. Als Hilfe kann folgendes Notebook dienen: https://www.kaggle.com/code/jhoward/is-it-a-bird-creating-a-model-from-your-own-data

Fragen

Was ist eine Convolution? Was sind Convolutional Neural Networks?
Was ist "Transfer Learning" bei neuronalen Netzwerken? Warum ist das hilfreich?
Was ist eine "Loss Function"? Wie funktioniert "Cross Entropy" (Kreuzentropie?)
Was ist Gradient Descent (oder Stochastic Gradient Descent)? Wofür brauchen wir das?

Röntgen Thorax (Lunge) Klassifikation mittels Deep Learning

https://colab.research.google.com/drive/105ulWzBwLaYCSB-4GXdTJ_9Yr68zi7uN

# CNN Image Classification Tutorial

## 1. Import Libraries

import torch
import torch.nn as nn
import torch.optim as optim
import torchvision
import torchvision.transforms as transforms
import matplotlib.pyplot as plt
import numpy as np

## 2. Load and Preprocess Data

# Define transformations
transform = transforms.Compose([
    transforms.ToTensor(),
    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])

# Load CIFAR-10 dataset
trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=transform)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=32, shuffle=True)

# Visualize some images
def imshow(img):
    img = img / 2 + 0.5     # unnormalize
    npimg = img.numpy()
    plt.imshow(np.transpose(npimg, (1, 2, 0)))
    plt.show()

# Get random training images
dataiter = iter(trainloader)
images, labels = next(dataiter)

# Show images
imshow(torchvision.utils.make_grid(images))

## 3. Define the CNN Architecture

class SimpleCNN(nn.Module):
    def __init__(self):
        super(SimpleCNN, self).__init__()
        self.conv1 = nn.Conv2d(3, 16, 3, padding=1)
        self.conv2 = nn.Conv2d(16, 32, 3, padding=1)
        self.pool = nn.MaxPool2d(2, 2)
        self.fc1 = nn.Linear(32 * 8 * 8, 128)
        self.fc2 = nn.Linear(128, 10)

    def forward(self, x):
        x = self.pool(torch.relu(self.conv1(x)))
        x = self.pool(torch.relu(self.conv2(x)))
        x = x.view(-1, 32 * 8 * 8)
        x = torch.relu(self.fc1(x))
        x = self.fc2(x)
        return x

# Instantiate the model
model = SimpleCNN()

## 4. Define Loss Function and Optimizer

criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=0.001, momentum=0.9)

## 5. Train the Model

num_epochs = 5
for epoch in range(num_epochs):
    running_loss = 0.0
    for i, data in enumerate(trainloader, 0):
        inputs, labels = data
        optimizer.zero_grad()
        outputs = model(inputs)
        loss = criterion(outputs, labels)
        loss.backward()
        optimizer.step()
        running_loss += loss.item()
        if i % 200 == 199:
            print(f'[{epoch + 1}, {i + 1}] loss: {running_loss / 200:.3f}')
            running_loss = 0.0

print('Finished Training')

## 6. Evaluate the Model

# Function to show images
def imshow(img):
    img = img / 2 + 0.5     # unnormalize
    npimg = img.numpy()
    plt.imshow(np.transpose(npimg, (1, 2, 0)))
    plt.show()

# Get random training images
dataiter = iter(trainloader)
images, labels = next(dataiter)

# Show images
imshow(torchvision.utils.make_grid(images))
print('GroundTruth: ', ' '.join(f'{trainset.classes[labels[j]]:5s}' for j in range(4)))

# Print predictions
outputs = model(images)
_, predicted = torch.max(outputs, 1)
print('Predicted: ', ' '.join(f'{trainset.classes[predicted[j]]:5s}' for j in range(4)))

## 7. Calculate Accuracy on the Entire Dataset

correct = 0
total = 0
with torch.no_grad():
    for data in trainloader:
        images, labels = data
        outputs = model(images)
        _, predicted = torch.max(outputs.data, 1)
        total += labels.size(0)
        correct += (predicted == labels).sum().item()

print(f'Accuracy of the network on the 50000 train images: {100 * correct // total} %')

## 8. Exercise for Students
# Try modifying the CNN architecture or training parameters to improve accuracy
# Hint: You could add more convolutional layers, change the number of filters, or adjust the learning rate

vacmar01/deep_learning.md

Allgemeine Tips

Links

Aufgabe

Fragen

Röntgen Thorax (Lunge) Klassifikation mittels Deep Learning