Zefs Guide to Deep Learning

Acknowledgments

1Introduction

1. Why deep learning?
2. Why this book?
3. What does this book cover and not cover?
4. How to use this book

2Machine Learning

1. What is machine learning?
2. Types of machine learning tasks and solutions
3. Regression
4. Classification
5. Supervised learning
6. Unsupervised learning
7. Self-supervised learning
8. Reinforcement learning
9. An example task
10. Predicting real estate sales prices
11. Formulating machine learning problems
12. Data sets and features
13. Measuring performance
14. Performance baselines and success thresholds
15. Model selection
16. Model training
17. Supervised learning
18. Unsupervised learning
19. Loss functions
20. Parameter optimization
21. Generalization and overfitting
22. Avoiding overfitting
23. Hyperparameters
24. Productionization
25. Common issues
26. Common machine learning models
27. From “traditional” ML to deep learning
28. References

3Neural Networks

1. What is a neural network?
2. What are some tasks that neural networks can accomplish?
3. The building blocks of neural networks
4. Activation functions
5. Neural network layers
6. Connections, weights, and biases
7. Learning via gradient descent
8. Output layers
9. What does a neural network do?
10. From basic neural networks to deep learning
11. Resources

4The rise of deep learning

1. Moving to deep neural networks
2. What made deep neural networks possible?
3. Where are we now with deep learning?

5Computer vision and convolutional neural networks

1. Computers and images
2. Computer vision tasks
3. Traditional computer vision
4. What’s hard about computer vision tasks?
5. Convolutional neural networks
6. Convolutions
7. Filter size, strides, padding, and pooling
8. A basic CNN architecture
9. Some important CNN model architectures for computer vision tasks
10. AlexNet
11. ResNet
12. U-Net for semantic segmentation
13. YOLO for object detection
14. Image generation with GANs
15. Common CNN techniques
16. Regularization
17. Data augmentation
18. Batch normalization
19. Gradient descent algorithms
20. Transfer learning
21. Summary and resources

6Natural language processing and sequential data techniques

1. Text, natural language, and sequential data
2. Types of sequential tasks
3. Traditional approaches
4. Making a neural network remember
5. The recurrent neural network
6. Creating context with embeddings
7. Embeddings
8. Architectures for sequential tasks
9. Gated recurrent units
10. Long short-term memory
11. Attention
12. Transformers
13. Applications and Transformer based architectures
14. Summary and resources

7Advanced techniques and practical considerations

1. Combining vision and language
2. Image captioning
3. Joint embeddings
4. Diffusion models
5. Self-supervised learning
6. Image-based techniques
7. Contrastive learning
8. Math topics related to deep learning
9. Linear algebra
10. Statistics and probability
11. Differential calculus
12. Machine learning engineering
13. Deep learning libraries
14. Graphical processing units and specialized hardware
15. Machine learning systems
16. Wrapping up