来自巴斯大学计算机科学教授Simon J.D. Prince撰写的《理解深度学习》新书，共有19章，从机器学习基础概念到深度学习各种模型，包括最新的Transformer和图神经网络，比较系统全面，值得关注。

Chapter 1 - 导论 Introduction Chapter 2 - 监督学习 Supervised learning Chapter 3 - 浅层神经网络 Shallow neural networks Chapter 4 - 深度神经网络 Deep neural networks Chapter 5 - 损失函数 Loss functions Chapter 6 - 训练模型 Training models Chapter 7 - 梯度与初始化 Gradients and initialization Chapter 8 - 度量性能 Measuring performance Chapter 9 - 正则化 Regularization Chapter 10 - 卷积网络 Convolutional nets Chapter 11 - 残差网络 Residual networks and BatchNorm Chapter 12 - Transformers Chapter 13 - 图神经网络 Graph neural networks Chapter 14 -变分自编码器 Variational auto-encoders Chapter 15 - Normalizing flows Chapter 16 - 生成对抗网络 Generative adversarial networks Chapter 17 - 扩散模型 Diffusion models Chapter 18 - 深度强化学习 Deep reinforcement learning Chapter 19 - 为什么深度学习有效 Why does deep learning work?