Hands-On One-shot Learning with Python: Learn to implement fast and accurate deep learning models with fewer training samples using PyTorch. Code

Cover
Title Page
Copyright and Credits
About Packt
Contributors
Table of Contents
Preface
Section 1: One-shot Learning Introduction
Chapter 1: Introduction to One-shot Learning
	Technical requirements
	The human brain – overview
		How the human brain learns
		Comparing human neurons and artificial neurons
	Machine learning – historical overview
		Challenges in machine learning and deep learning
	One-shot learning – overview
		Prerequisites of one-shot learning
		Types of one-shot learning
	Setting up your environment
	Coding exercise
		kNN – basic one-shot learning
	Summary
	Questions
Section 2: Deep Learning Architectures
Chapter 2: Metrics-Based Methods
	Technical requirements
	Parametric methods – an overview
		Neural networks – learning procedure
		Visualizing parameters
	Understanding Siamese networks
		Architecture
		Preprocessing
		Contrastive loss function
		Triplet loss function
			Applications
	Understanding matching networks
		Model architecture
			Training procedure
			Modeling level – the matching networks architecture
	Coding exercise
		Siamese networks – the MNIST dataset
		Matching networks – the Omniglot dataset
	Summary
	Questions
	Further reading
Chapter 3: Model-Based Methods
	Technical requirements
	Understanding Neural Turing Machines
		Architecture of an NTM
		Modeling 
			Reading 
			Writing
			Addressing
	Memory-augmented neural networks
		Reading
		Writing
	Understanding meta networks
		Algorithm of meta networks
			Algorithm
	Coding exercises
		Implementation of NTM
		Implementation of MAAN
	Summary
	Questions
	Further reading
Chapter 4: Optimization-Based Methods
	Technical requirements
	Overview of gradient descent
	Understanding model-agnostic meta-learning
		Understanding the logic behind MAML
			Algorithm
		MAML application – domain-adaptive meta-learning
	Understanding LSTM meta-learner
		Architecture of the LSTM meta-learner
			Data preprocessing
			Algorithm – pseudocode implementation
	Exercises
		A simple implementation of model-agnostic meta-learning
		A simple implementation of domain-adaption meta-learning
	Summary
	Questions
	Further reading
Section 3: Other Methods and Conclusion
Chapter 5: Generative Modeling-Based Methods
	Technical requirements
	Overview of Bayesian learning
	Understanding directed graphical models
	Overview of probabilistic methods
	Bayesian program learning
		Model
			Type generation
			Token generation
			Image generation
	Discriminative k-shot learning
		Representational learning
		Probabilistic model of the weights
			Choosing a model for the weights
		Computation and approximation for each phase 
			Phase 1 – representation learning
			Phase 2 – concept learning
			Phase 3 – k-shot learning
			Phase 4 – k-shot testing
	Summary
	Further reading
Chapter 6: Conclusions and Other Approaches
	Recent advancements
		Object detection in few-shot domains
		Image segmentation in few-shot domains
	Related fields
		Semi-supervised learning
		Imbalanced learning 
		Meta-learning
		Transfer learning
	Applications
	Further reading
Other Books You May Enjoy
Index