Programming Machine Learning Coding-From Deep Learning

《Programming Machine Learning》这本书是为开发者编写的机器学习入门书籍，旨在帮助开发者理解机器学习的基本原理，并能够编写自己的机器学习程序。书中的例子使用Python语言，从最基础的线性回归开始，逐步引导读者构建复杂的机器学习系统。内容包括监督学习、神经网络和深度学习，并且所有代码都是从零开始编写的，以帮助读者深入理解每一行代码的含义。

目录

Early Praise for Programming Machine Learning

• 引言部分，包括多个对本书的高度评价。

Contents

• Acknowledgments (致谢)
• How the Heck Is That Possible? (机器学习的神奇之处)

Part I — From Zero to Image Recognition

1. How Machine Learning Works (机器学习的工作原理)

   • Programming vs. Machine Learning (编程与机器学习的对比)
   • Supervised Learning (监督学习)
   • The Math Behind the Magic (背后的数学原理)
   • Setting Up Your System (系统设置)

2. Your First Learning Program (你的第一个学习程序)

   • Getting to Know the Problem (了解问题)
   • Coding Linear Regression (编写线性回归代码)
   • Adding a Bias (添加偏差)

3. Walking the Gradient (梯度下降)

   • Our Algorithm Doesn’t Cut It (我们的算法不够好)
   • Gradient Descent (梯度下降)
   • Hands On: Basecamp Overshooting (实践：基准站点过冲)

4. Hyperspace! (高维空间)

   • Adding More Dimensions (添加更多维度)
   • Matrix Math (矩阵数学)
   • Upgrading the Learner (升级学习器)
   • Hands On: Field Statistician (实践：现场统计员)

5. A Discerning Machine (辨别机器)

   • Where Linear Regression Fails (线性回归的局限)
   • Invasion of the Sigmoids (S形曲线的引入)
   • Classification in Action (分类实践)

6. Getting Real (走向现实)

   • Data Come First (数据优先)
   • Our Own MNIST Library (我们的MNIST库)
   • The Real Thing (实际操作)

7. The Final Challenge (最后的挑战)

   • Going Multiclass (多类分类)
   • Moment of Truth (真相时刻)

8. The Perceptron (感知器)

   • Enter the Perceptron (引入感知器)
   • Assembling Perceptrons (组装感知器)
   • Where Perceptrons Fail (感知器的局限)

Part II — Neural Networks

1. Designing the Network (设计网络)

   • Assembling a Neural Network from Perceptrons (用感知器组装神经网络)
   • Enter the Softmax (引入Softmax)
   • Hands On: Network Adventures (实践：网络冒险)

2. Building the Network (构建网络)

   • Coding Forward Propagation (编码前向传播)
   • Cross Entropy (交叉熵)
   • Hands On: Time Travel Testing (实践：时光旅行测试)

3. Training the Network (训练网络)

   • The Case for Backpropagation (反向传播的必要性)
   • Applying Backpropagation (应用反向传播)

4. How Classifiers Work (分类器的工作原理)

   • Tracing a Boundary (追踪边界)
   • Bending the Boundary (弯曲边界)

5. Batchin’ Up (批量处理)

   • Learning, Visualized (可视化学习)
   • Batch by Batch (逐批处理)

6. The Zen of Testing (测试的禅意)

   • The Threat of Overfitting (过拟合的威胁)
   • A Testing Conundrum (测试难题)

7. Let’s Do Development (开始开发)

   • Preparing Data (准备数据)
   • Tuning Hyperparameters (调整超参数)

Part III — Deep Learning

1. A Deeper Kind of Network (更深层的网络)

   • Building a Neural Network with Keras (用Keras构建神经网络)
   • Making It Deep (使其变深)

2. Defeating Overfitting (战胜过拟合)

   • Overfitting Explained (解释过拟合)
   • Regularizing the Model (正则化模型)

3. Taming Deep Networks (驯服深度网络)

   • Understanding Activation Functions (理解激活函数)
   • Adding More Tricks to Your Bag (添加更多技巧)

4. Beyond Vanilla Networks (超越基本网络)

   • The CIFAR-10 Dataset (CIFAR-10数据集)
   • The Building Blocks of CNNs (卷积神经网络的构建块)

5. Into the Deep (深入探索)

   • The Rise of Deep Learning (深度学习的崛起)
   • Unreasonable Effectiveness (非凡的效果)

Appendices

A1. Just Enough Python (足够的Python) - What Python Looks Like (Python的样子) - Python’s Building Blocks (Python的构建块)

A2. The Words of Machine Learning (机器学习的术语) - Index (索引)