Python Deep Learning Cookbook

Contents......Page 3
Preface......Page 9
Programming Environments, GPU Computing, Cloud Solutions & Deep Learning Frameworks......Page 14
Setting up a deep learning environment......Page 15
Launching an instance on Amazon Web Services (AWS)......Page 16
Launching an instance on Google Cloud Platform (GCP)......Page 17
Installing CUDA and cuDNN......Page 18
Installing Anaconda and libraries......Page 21
Connecting with Jupyter Notebooks on a server......Page 22
Building state-of-the-art, production-ready models with TensorFlow......Page 23
Intuitively building networks with Keras......Page 26
Using PyTorch’s dynamic computation graphs for RNNs......Page 28
Implementing high-performance models with CNTK......Page 30
Building efficient models with MXNet......Page 32
Defining networks using simple and efficient code with Gluon......Page 34
Introduction......Page 36
Understanding the perceptron......Page 37
Implementing a single-layer neural network......Page 41
Building a multi-layer neural network......Page 45
Getting started with activation functions......Page 49
Experiment with hidden layers and hidden units......Page 55
Implementing an autoencoder......Page 59
Tuning the loss function......Page 62
Experimenting with different optimizers......Page 65
Improving generalization with regularization......Page 68
Adding dropout to prevent overfitting......Page 73
Introduction......Page 79
Applying pooling layers......Page 84
Optimizing with batch normalization......Page 86
Understanding padding and strides......Page 91
Experimenting with different types of initialization......Page 97
Implementing a convolutional autoencoder......Page 101
Applying a 1D CNN to text......Page 104
Introduction......Page 107
Implementing a simple RNN......Page 108
Adding Long Short-Term Memory (LSTM)......Page 111
Using gated recurrent units (GRUs)......Page 114
Implementing bidirectional RNNs......Page 116
Character-level text generation......Page 118
Introduction......Page 122
Implementing policy gradients......Page 123
Implementing a deep Q-learning algorithm......Page 131
Introduction......Page 140
Understanding GANs......Page 141
Implementing Deep Convolutional GANs (DCGANs)......Page 144
Upscaling the resolution of images with Super-Resolution GANs (SRGANs)......Page 150
Introduction......Page 158
Augmenting images with computer vision techniques......Page 159
Classifying objects in images......Page 164
Localizing an object in images......Page 169
Segmenting classes in images with U-net......Page 175
Scene understanding (semantic segmentation)......Page 180
Finding facial key points......Page 183
Recognizing faces......Page 188
Transferring styles to images......Page 195
Analyzing sentiment......Page 201
Translating sentences......Page 204
Summarizing text......Page 209
Introduction......Page 214
Implementing a speech recognition pipeline from scratch......Page 215
Identifying speakers with voice recognition......Page 218
Understanding videos with deep learning......Page 222
Introduction......Page 226
Predicting stock prices with neural networks......Page 227
Predicting bike sharing demand......Page 231
Using a shallow neural network for binary classification......Page 235
Introduction......Page 237
Learning to drive a car with end-to-end learning......Page 238
Learning to play games with deep reinforcement learning......Page 242
Genetic Algorithm (GA) to optimize hyperparameters......Page 250
Hyperparameter Selection, Tuning & Neural Network Learning......Page 256
Visualizing training with TensorBoard and Keras......Page 257
Working with batches and mini-batches......Page 262
Using grid search for parameter tuning......Page 266
Learning rates and learning rate schedulers......Page 269
Comparing optimizers......Page 272
Determining the depth of the network......Page 275
Adding dropouts to prevent overfitting......Page 276
Making a model more robust with data augmentation......Page 281
Visualizing training with TensorBoard......Page 284
Analyzing network weights and more......Page 289
Freezing layers......Page 294
Storing the network topology and trained weights......Page 296
Introduction......Page 300
Large-scale visual recognition with GoogLeNet/Inception......Page 301
Extracting bottleneck features with ResNet......Page 303
Leveraging pretrained VGG models for new classes......Page 304
Fine-tuning with Xception......Page 307
Index......Page 311