Intelligent Projects Using Python
Santanu Pattanayak更新时间：2021-07-02 14:11:25
最新章节：Leave a review - let other readers know what you thinkcoverpage
Title Page
Copyright and Credits
Intelligent Projects Using Python
About Packt
Why subscribe?
Packt.com
Contributors
About the author
About the reviewer
Packt is searching for authors like you
Preface
Who this book is for
What this book covers
To get the most out of this book
Download the example code files
Download the color images
Code in action
Conventions used
Get in touch
Reviews
Foundations of Artificial Intelligence Based Systems
Neural networks
Neural activation units
Linear activation units
Sigmoid activation units
The hyperbolic tangent activation function
Rectified linear unit (ReLU)
The softmax activation unit
The backpropagation method of training neural networks
Convolutional neural networks
Recurrent neural networks (RNNs)
Long short-term memory (LSTM) cells
Generative adversarial networks
Reinforcement learning
Q-learning
Deep Q-learning
Transfer learning
Restricted Boltzmann machines
Autoencoders
Summary
Transfer Learning
Technical requirements
Introduction to transfer learning
Transfer learning and detecting diabetic retinopathy
The diabetic retinopathy dataset
Formulating the loss function
Taking class imbalances into account
Preprocessing the images
Additional data generation using affine transformation
Rotation
Translation
Scaling
Reflection
Additional image generation through affine transformation
Network architecture
The VGG16 transfer learning network
The InceptionV3 transfer learning network
The ResNet50 transfer learning network
The optimizer and initial learning rate
Cross-validation
Model checkpoints based on validation log loss
Python implementation of the training process
Dynamic mini batch creation during training
Results from the categorical classification
Inference at testing time
Performing regression instead of categorical classification
Using the keras sequential utils as generator
Summary
Neural Machine Translation
Technical requirements
Rule-based machine translation
The analysis phase
Lexical transfer phase
Generation phase
Statistical machine-learning systems
Language model
Perplexity for language models
Translation model
Neural machine translation
The encoder–decoder model
Inference using the encoder–decoder model
Implementing a sequence-to-sequence neural translation machine
Processing the input data
Defining a model for neural machine translation
Loss function for the neural translation machine
Training the model
Building the inference model
Word vector embeddings
Embeddings layer
Implementing the embeddings-based NMT
Summary
Style Transfer in Fashion Industry using GANs
Technical requirements
DiscoGAN
CycleGAN
Learning to generate natural handbags from sketched outlines
Preprocess the Images
The generators of the DiscoGAN
The discriminators of the DiscoGAN
Building the network and defining the cost functions
Building the training process
Important parameter values for GAN training
Invoking the training
Monitoring the generator and the discriminator loss
Sample images generated by DiscoGAN
Summary
Video Captioning Application
Technical requirements
CNNs and LSTMs in video captioning
A sequence-to-sequence video-captioning system
Data for the video-captioning system
Processing video images to create CNN features
Processing the labelled captions of the video
Building the train and test dataset
Building the model
Definition of the model variables
Encoding stage
Decoding stage
Building the loss for each mini-batch
Creating a word vocabulary for the captions
Training the model
Training results
Inference with unseen test videos
Inference function
Results from evaluation
Summary
The Intelligent Recommender System
Technical requirements
What is a recommender system?
Latent factorization-based recommendation system
Deep learning for latent factor collaborative filtering
The deep learning-based latent factor model
SVD++
Training model with SVD++ on the Movie Lens 100k dataset
Restricted Boltzmann machines for recommendation
Contrastive divergence
Collaborative filtering using RBMs
Collaborative filtering implementation using RBM
Processing the input
Building the RBM network for collaborative filtering
Training the RBM
Inference using the trained RBM
Summary
Mobile App for Movie Review Sentiment Analysis
Technical requirements
Building an Android mobile app using TensorFlow mobile
Movie review rating in an Android app
Preprocessing the movie review text
Building the model
Training the model
The batch generator
Freezing the model to a protobuf format
Creating a word-to-token dictionary for inference
App interface page design
The core logic of the Android app
Testing the mobile app
Summary
Conversational AI Chatbots for Customer Service
Technical requirements
Chatbot architecture
A sequence-to-sequence model using an LSTM
Building a sequence-to-sequence model
Customer support on Twitter
Creating data for training the chatbot
Tokenizing the text into word indices
Replacing anonymized screen names
Defining the model
Loss function for training the model
Training the model
Generating output responses from the model
Putting it all together
Invoking the training
Results of inference on some input tweets
Summary
Autonomous Self-Driving Car Through Reinforcement Learning
Technical requirements
Markov decision process
Learning the Q value function
Deep Q learning
Formulating the cost function
Double deep Q learning
Implementing an autonomous self-driving car
Discretizing actions for deep Q learning
Implementing the Double Deep Q network
Designing the agent
The environment for the self-driving car
Putting it all together
Helper functions
Results from the training
Summary
CAPTCHA from a Deep-Learning Perspective
Technical requirements
Breaking CAPTCHAs with deep learning
Generating basic CAPTCHAs
Generating data for training a CAPTCHA breaker
Captcha breaker CNN architecture
Pre-processing the CAPTCHA images
Converting the CAPTCHA characters to classes
Data generator
Training the CAPTCHA breaker
Accuracy on the test data set
CAPTCHA generation through adversarial learning
Optimizing the GAN loss
Generator network
Discriminator network
Training the GAN
Noise distribution
Data preprocessing
Invoking the training
The quality of CAPTCHAs during training
Using the trained generator to create CAPTCHAs for use
Summary
Other Books You May Enjoy
Leave a review - let other readers know what you think
更新时间：2021-07-02 14:11:25