深度学习基础

Preface
1. The Neural Network
Building Intelligent Machines
The Limits of Traditional Computer Programs
The Mechanics of Machine Learning
The Neuron
Expressing Linear Perceptrons as Neurons
Feed-Forward Neural Networks
Linear Neurons and Their Limitations
Sigmoid， Tanh， and ReLU Neurons
Softmax Output Layers
Looking Forward

2. Training Feed-Forward Neural Networks
The Fast-Food Problem
Gradient Descent
The Delta Rule and Learning Rates
Gradient Descent with Sigmoidal Neurons
The BackpropagatioAlgorithm
Stochastic and Minibatch Gradient Descent
Test Sets， ValidatioSets， and Overfitting
Preventing Overfitting iDeep Neural Networks
Summary

3. Implementing Neural Networks iTensorFIow
What Is TensorFlow
How Does TensorFlow Compare to Alternatives
Installing TensorFlow
Creating and Manipulating TensorFlow Variables
TensorFlow Operations
Placeholder Tensors
Sessions iTensorFlow
Navigating Variable Scopes and Sharing Variables
Managing Models over the CPU and GPU
Specifying the Logistic RegressioModel iTensorFlow
Logging and Training the Logistic RegressioModel
Leveraging TensorBoard to Visualize ComputatioGraphs and Learning
Building a Multilayer Model for MNIST iTensorFlow
Summary

4. Beyond Gradient Descent
The Challenges with Gradient Descent
Local Minima ithe Error Surfaces of Deep Networks
Model Identifiability
How Pesky Are Spurious Local Minima iDeep Networks
Flat Regions ithe Error Surface
Whethe Gradient Points ithe Wrong Direction
Momentum-Based Optimization
A Brief View of Second-Order Methods
Learning Rate Adaptation
AdaGrad——Accumulating Historical Gradients
RMSProp——Exponentially Weighted Moving Average of Gradients
Adam——Combining Momentum and RMSProp
The Philosophy Behind Optimizer Selection
Summary

5. Convolutional Neural Networks
Neurons iHumaVision
The Shortings of Feature Selection
Vanilla Deep Neural Networks Don't Scale
Filters and Feature Maps
Full Descriptioof the Convolutional Layer
Max Pooling
Full Architectural Descriptioof ConvolutioNetworks
Closing the Loop oMNIST with Convolutional Networks
Image Preprocessing Pipelines Enable More Robust Models
Accelerating Training with Batch Normalization
Building a Convolutional Network for CIFAR-10
Visualizing Learning iConvolutional Networks
Leveraging Convolutional Filters to Replicate Artistic Styles
Learning Convolutional Filters for Other Problem Domains
Summary

6. Embedding and RepresentatioLearning
Learning Lower-Dimensional Representations
Principal Component Analysis
Motivating the Autoencoder Architecture
Implementing aAutoencoder iTensorFlow
Denoising to Force Robust Representations
Sparsity iAutoencoders
WheContext Is More Informative thathe Input Vector
The Word2Vec Framework
Implementing the Skip-Gram Architecture
Summary

7. Models for Sequence Analysis
Analyzing Variable-Length Inputs
Tackling seq2seq with Neural N-Grams
Implementing a Part-of-Speech Tagger
Dependency Parsing and SyntaxNet
Beam Search and Global Normalization
A Case for Stateful Deep Learning Models
Recurrent Neural Networks
The Challenges with Vanishing Gradients
Long Short-Term Memory （LSTM） Units
TensorFlow Primitives for RNN Models
Implementing a Sentiment Analysis Model
Solving seq2seq Tasks with Recurrent Neural Networks
Augmenting Recurrent Networks with Attention
Dissecting a Neural TranslatioNetwork
Summary

8. Memory Augmented Neural Networks
Neural Turing Machines
Attention-Based Memory Access
NTM Memory Addressing Mechanisms
Differentiable Neural Computers
Interference-Free Writing iDNCs
DNC Memory Reuse
Temporal Linking of DNC Writes
Understanding the DNC Read Head
The DNC Controller Network
Visualizing the DNC iAction
Implementing the DNC iTensorFlow
Teaching a DNC to Read and Comprehend
Summary

9. Deep Reinforcement Learning
Deep Reinforcement Learning Masters Atari Games
What Is Reinforcement Learning
Markov DecisioProcesses （MDP）
Policy
Future Return
Discounted Future Return
Explore Versus Exploit
Policy Versus Value Learning
Policy Learning via Policy Gradients
Pole-Cart with Policy Gradients
OpenAI Gym
Creating aAgent
Building the Model and Optimizer
Sampling Actions
Keeping Track of History
Policy Gradient MaiFunction
PGAgent Performance oPole-Cart
Q-Learning and Deep Q-Networks
The BellmaEquation
Issues with Value Iteration
Approximating the Q-Function
Deep Q-Network （DQN）
Training DQN
Learning Stability
Target Q-Network
Experience Replay
From Q-Functioto Policy
DQN and the Markov Assumption
DQN's Solutioto the Markov Assumption
Playing Breakout wth DQN
Building Our Architecture
Stacking Frames
Setting Up Training Operations
Updating Our Target Q-Network
Implementing Experience Replay
DQN MaiLoop
DQNAgent Results oBreakout
Improving and Moving Beyond DQN
Deep Recurrent Q-Networks （DRQN）
Asynchronous Advantage Actor-Critic Agent （A3C）
UNsupervised REinforcement and Auxiliary Learning （UNREAL）
Summary
Index