Principles and Labs for Deep Learning

Principles and Labs for Deep Learning provides the knowledge and techniques needed to help readers design and develop deep learning models. Deep Learning techniques are introduced through theory, comprehensively illustrated, explained through the TensorFlow source code examples, and analyzed through the visualization of results. The structured methods and labs provided by Dr. Huang and Dr. Le enable readers to become proficient in TensorFlow to build deep Convolutional Neural Networks (CNNs) through custom APIs, high-level Keras APIs, Keras Applications, and TensorFlow Hub. Each chapter has one corresponding Lab with step-by-step instruction to help the reader practice and accomplish a specific learning outcome.

Deep Learning has been successfully applied in diverse fields such as computer vision, audio processing, robotics, natural language processing, bioinformatics and chemistry. Because of the huge scope of knowledge in Deep Learning, a lot of time is required to understand and deploy useful, working applications, hence the importance of this new resource. Both theory lessons and experiments are included in each chapter to introduce the techniques and provide source code examples to practice using them. All Labs for this book are placed on GitHub to facilitate the download. The book is written based on the assumption that the reader knows basic Python for programming and basic Machine Learning.

• Introduces readers to the usefulness of neural networks and Deep Learning methods
• Provides readers with in-depth understanding of the architecture and operation of Deep Convolutional Neural Networks
• Demonstrates the visualization needed for designing neural networks
• Provides readers with an in-depth understanding of regression problems, binary classification problems, multi-category classification problems, Variational Auto-Encoder, Generative Adversarial Network, and Object detection

Scientists, researchers, and MSc. PhD. students from the fields of Computer Science, Artificial Intelligence, and Information Technology. The audience includes researchers and practitioners in any field who needs to learn the practical steps for developing and applying Deep Learning to a variety of applications

Chapter 0: Environment Installation
0.1. Python Installation
0.2. TensorFlow Installation
0.3. Python extension installation
0.4. Jupyter Notebook
0.5. GitHub code
0.6. PyCharm IDE

Chapter 1: Introduction to TensorFlow2.0
1.1. What is Deep Learning
1.2. Create a new project
1.3. Introduction to TensorFlow
1.4. Eager Execution
1.4.1. Introduction to Eager Execution
1.4.2. Basic operations
1.5. Keras
1.5.1. Introduction to Keras
1.5.2. Sequential Model
1.5.3. Functional API
1.6. tf.data
1.6.1. Introduction to tf.data
1.6.2. Basic Operations

Chapter 2: Regression Problem
2.1. Deep Neural Network
2.1.1. A brief history of neural networks
2.1.2. Principle of Neural Network
2.1.3. Fully connected
2.1.4. MSE and MAE Loss functions
2.1.5. Neural network weight update
2.1.6. Neural network training steps
2.2. Introduction to Kaggle
2.3. Experiment 1: House price prediction model
2.3.1. Introduction to the dataset
2.3.2. Create Project
2.3.3. Source code
2.4. Introduction to TensorBoard
2.5. Experiment 2: Overfitting
2.5.1. Introduction to the dataset
2.5.2. Create Project
2.5.3. Source code

Chapter 3: Binary classification problem
3.1. Machine Learning Algorithms
3.2. Binary Classification Problem
3.2.1. Logistic Regression
3.2.2. Sigmoid
3.2.3. Binary Cross-Entropy
3.2.4. One-hot Encoding
3.3. Experiment: Pokémon Battle Prediction
3.3.1. Introduction to the dataset
3.3.2. Create Project
3.3.3. Source code

Chapter 4: Multi-category Classification Problem
4.1. Convolutional Neural Network
4.1.1. Introduction to Convolutional Neural Network
4.1.2. The architecture of Convolutional Neural Network
4.1.3. Principle of Convolutional Neural Network
4.2. Multi-category Classification Problem
4.2.1. Softmax
4.2.2. Categorical Cross-Entropy
4.2.3. Data enhancement
4.3. Experiment: CIFAR-10 image recognition
4.3.1. Introduction to the dataset
4.3.2. TensorFlow Datasets
4.3.3. Create Project
4.3.4. Source code

Chapter 5: Training Neural Network
5.1. Backpropagation
5.2. Weight Initialization
5.2.1. Normal Distribution
5.2.2. Xavier/Glorot initialization
5.2.3. He initialization
5.3. Batch Normalization
5.3.1. Introduction to Batch Normalization
5.3.2. Neural Network with Batch Normalization
5.4. Experiment 1: Using CIFAR-10 dataset to verify three weight initialization methods
5.5. Experiment 2: Using CIFAR-10 dataset to verify Batch Normalization method
5.6. Performance Comparison of Different Neural Networks

Chapter 6: Advanced TensorFlow2.0
6.1. Advanced TensorFlow
6.1.1. Customized network layer
6.1.2. Customized loss function
6.1.3. Customized metric function
6.1.4. Customized callback function
6.2. Comparison of high-level Keras API and customized API
6.2.1. Network Layer
6.2.2. Loss Function
6.2.3. Metric Function
6.2.4. Callback Function
6.3. Experiment: Implementation of two network models using high-level Keras API and Customized API
6.3.1. Create Project
6.3.2. Source Code

Chapter 7: Advanced TensorBoard
7.1. Advanced TensorBoard
7.1.1. tf.summary
7.1.2. tf.summary.scalar
7.1.3. tf.summary.image
7.1.4. tf.summary.text
7.1.5. tf.summary.audio
7.1.6. tf.summary.histogram
7.2. Experiment 1: Using tf.summary.image to visualize training results
7.2.1. Create a project
7.2.2. Source Code
7.3. Experiment 2: Using hyperparameter tuning with TensorBoard HParams to analyze multiple training models 7.3.1. Open TensorBoard
7.3.2. Source Code

Chapter 8: Convolutional Neural Network Architectures
8.1. Most Popular Convolutional Neural Network Architectures
8.1.1. LeNet
8.1.2. AlexNet
8.1.3. VGG
8.1.4. GoogLeNet
8.1.5. ResNet
8.1.6. Comparison of network architectures
8.2. Experiment: Implementation of InceptionV3 network architecture
8.2.1. Create a project
8.2.2. Keras Applications
8.2.3. TensorFlow Hub

Chapter 9: Transfer Learning
9.1. Transfer Learning
9.1.1. Introduction to Transfer Learning
9.1.2. Transfer learning methods
9.2. Experiment: Using Inception V3 for Transfer Learning
9.2.1. Create Project
9.2.2. Introduction to the Dataset
9.2.3. Source Code

Chapter 10: Variational Auto-Encoder
10.1. Introduction to Auto-Encoder
10.2. Introduction to Variational Auto-Encoder
10.3. Variational Auto-Encoder loss function
10.4. Experiment: Implementation of Variational Auto-Encoder Model
10.4.1. Create Project
10.4.2. Introduction to Dataset
10.4.3. Building Variational Auto-Encoder model.
10.4.4. Training Variational Auto-Encoder model

Chapter 11: WGAN-GP
11.1. Generative Adversarial Network
11.1.1. Introduction to Generative Adversarial Network
11.1.2. Generative Adversarial Network Training and Loss Function
11.2. Evaluation of GAN, WGAN, and WGAN-GP
11.2.1. Problems of Generative Adversarial Network
11.2.2. Introduction to Wasserstein distance
11.2.3. WGAN-GP Loss Function
11.3. Experiment：Implementation of WGAN-GP
11.3.1. Create Project
11.3.2. Introduction to Dataset
11.3.3. Explanation of WGAN-GP Project

Chapter 12: Object Detection
12.1. Computer Vision
12.2. Introduction to Object Detection
12.3. Historical Progress of Object Detection
12.3.1. R-CNN
12.3.2. Fast R-CNN
12.3.3. Faster R-CNN
12.3.4. YOLO v1
12.3.5. SSD
12.3.6. YOLO v2
12.3.7. FPN
12.3.8. RetinaNet
12.3.9. YOLO v3 12.3.10. CFF-SSD
12.3.11. DSNet
12.4. Experiment: Implementation of YOLO v3
12.4.1. Load project
12.4.2. Introduction to Dataset
12.4.3. Explanation of YOLO v3 project
12.4.4. YOLO v3 Training and Testing