Deep Learning from Scratch with PyTorch

Part II · Chapter 04 · 6 sections · 123 min

Mastering PyTorch's core abstractions

1. 4.1Tensors: The Foundation of Deep Learning25m
2. 4.2Tensor Operations Deep Dive22m
3. 4.3Indexing, Slicing, and Reshaping18m
4. 4.4GPU Computing15m
5. 4.5Autograd System25m
6. 4.6Building Custom Autograd Functions18m

Part II · Chapter 05 · 6 sections · 114 min

Components of neural networks

1. 5.1The nn.Module Class20m
2. 5.2Linear Layers15m
3. 5.3Activation Functions22m
4. 5.4Loss Functions20m
5. 5.5Normalization Layers22m
6. 5.6Dropout and Regularization15m

Part II · Chapter 06 · 6 sections · 118 min

Understanding neural network architectures from shallow to deep

1. 6.1The Perceptron18m
2. 6.2Multi-Layer Perceptrons (MLPs)22m
3. 6.3Universal Approximation Theorem20m
4. 6.4Shallow vs Deep Networks18m
5. 6.5Why Depth Matters15m
6. 6.6Building Your First Neural Network25m

Part II · Chapter 07 · 5 sections · 89 min

Efficient data pipelines

1. 7.1The Dataset Class18m
2. 7.2DataLoader Deep Dive20m
3. 7.3Data Transforms18m
4. 7.4Custom Datasets18m
5. 7.5Handling Imbalanced Data15m

Part II · Chapter 08 · 6 sections · 130 min

The algorithm that makes deep learning possible

1. 8.1The Learning Problem12m
2. 8.2Gradient Descent Variants18m
3. 8.3Computational Graphs15m
4. 8.4Backpropagation Derivation30m
5. 8.5Implementing Backprop from Scratch35m
6. 8.6Gradient Flow Analysis20m

Part II · Chapter 09 · 8 sections · 161 min

The art and science of training

1. 9.1The Training Loop18m
2. 9.2Optimizers25m
3. 9.3Learning Rate Strategies22m
4. 9.4Weight Initialization18m
5. 9.5Overfitting and Regularization20m
6. 9.6Hyperparameter Tuning18m
7. 9.7Debugging Neural Networks22m
8. 9.8Transfer Learning Fundamentals18m

Part III · Chapter 10 · 6 sections · 141 min

The convolution operation explained

1. 10.1Motivation for Convolutions18m
2. 10.2The Convolution Operation22m
3. 10.3Convolution Parameters20m
4. 10.4Implementing Convolution28m
5. 10.5Pooling Operations25m
6. 10.6Transposed Convolutions28m

Part III · Chapter 11 · 1 section · 32 min

Evolution of CNN architectures

1. 11.1U-Net for Semantic Segmentation32m

Part III · Chapter 12 · 3 sections · 57 min

Practical CNN techniques

1. 12.1Data Augmentation22m
2. 12.2Transfer Learning20m
3. 12.3Using Pretrained Models15m

Part IV · Chapter 15 · 3 sections · 68 min

Gated recurrent architectures

1. 15.1The Vanishing Gradient Problem15m
2. 15.2Long Short-Term Memory28m
3. 15.3Implementing LSTM25m

Part V · Chapter 19 · 1 section · 18 min

Learning on graph-structured data

1. 19.1Introduction to Graphs18m

Part VII · Chapter 24 · 4 sections · 67 min

Learning without labels

1. 24.1Why Self-Supervised Learning?15m
2. 24.2Pretext Tasks for Images22m
3. 24.3Pretext Tasks for Text18m
4. 24.4Pretext Tasks for Sequences12m