麻省理工出版 2023年最新深度学习综述手册

这本手册详细介绍了深度学习的各个方面，从基础的监督学习、不监督学习和强化学习，到深度学习模型的训练、性能测量和改进。手册还包括了特定架构如卷积网络、残差网络和Transformer的详细讨论，以及生成对抗网络、变分自编码器、正则化流和扩散模型等现代深度生成模型。此外，手册还探讨了深度学习的伦理问题，强调了技术进步带来的潜在风险和责任。

目录

        1.Introduction

• 1.1 Supervised learning

• 1.2 Unsupervised learning

• 1.3 Reinforcement learning

• 1.4 Ethics

• 1.5 Structure of book

• 1.6 Other books

• 1.7 How to read this book

        2.Supervised learning

• 2.1 Supervised learning overview

• 2.2 Linear regression example

• 2.3 Summary

  3.Shallow neural networks

• 3.1 Neural network example

• 3.2 Universal approximation theorem

• 3.3 Multivariate inputs and outputs

• 3.4 Shallow neural networks: general case

• 3.5 Terminology

• 3.6 Summary

        4.Deep neural networks

• 4.1 Composing neural networks

• 4.2 From composing networks to deep networks

• 4.3 Deep neural networks

• 4.4 Matrix notation

• 4.5 Shallow vs. deep neural networks

• 4.6 Summary

        5.Loss functions

• 5.1 Maximum likelihood

• 5.2 Recipe for constructing loss functions

• 5.3 Example 1: univariate regression

• 5.4 Example