Natural language processing with TensorFlow : teach language to machines using Python's deep learning library
| 000 | 00000nam u2200205 a 4500 | |
| 001 | 000045983660 | |
| 005 | 20190611160924 | |
| 008 | 190515s2018 enka b 001 0 eng d | |
| 020 | ▼a 9781788478311 | |
| 040 | ▼a 211009 ▼c 211009 ▼d 211009 | |
| 082 | 0 4 | ▼a 006.35 ▼2 23 |
| 084 | ▼a 006.35 ▼2 DDCK | |
| 090 | ▼a 006.35 ▼b G196n | |
| 100 | 1 | ▼a Ganegedara, Thushan. |
| 245 | 1 0 | ▼a Natural language processing with TensorFlow : ▼b teach language to machines using Python's deep learning library / ▼c Thushan Ganegedara. |
| 260 | ▼a Birmingham : ▼b Packt Publishing Ltd, ▼c c2018. | |
| 300 | ▼a xviii, 446 p. : ▼b ill. ; ▼c 24 cm. | |
| 504 | ▼a Includes bibliographical references and index. | |
| 650 | 0 | ▼a Natural language processing (Computer science). |
| 650 | 0 | ▼a Machine learning. |
| 650 | 0 | ▼a Python (Computer program language). |
| 776 | 0 8 | ▼i Online version: ▼a Ganegedara, Thushan. ▼t Natural language processing with TensorFlow : teach language to machines using Python's deep learning library ▼z 9781788477758 ▼w (211009) 000045985744 |
| 945 | ▼a KLPA |
소장정보
| No. | 소장처 | 청구기호 | 등록번호 | 도서상태 | 반납예정일 | 예약 | 서비스 |
|---|---|---|---|---|---|---|---|
| No. 1 | 소장처 중앙도서관/서고6층/ | 청구기호 006.35 G196n | 등록번호 111809469 (3회 대출) | 도서상태 대출가능 | 반납예정일 | 예약 | 서비스 |
컨텐츠정보
책소개
Write modern natural language processing applications using deep learning algorithms and TensorFlow
Key Features:
- Focuses on more efficient natural language processing using TensorFlow
- Covers NLP as a field in its own right to improve understanding for choosing TensorFlow tools and other deep learning approaches
- Provides choices for how to process and evaluate large unstructured text datasets
- Learn to apply the TensorFlow toolbox to specific tasks in the most interesting field in artificial intelligence
Book Description:
Natural language processing (NLP) supplies the majority of data available to deep learning applications, while TensorFlow is the most important deep learning framework currently available. Natural Language Processing with TensorFlow brings TensorFlow and NLP together to give you invaluable tools to work with the immense volume of unstructured data in today's data streams, and apply these tools to specific NLP tasks.
Thushan Ganegedara starts by giving you a grounding in NLP and TensorFlow basics. You'll then learn how to use Word2vec, including advanced extensions, to create word embeddings that turn sequences of words into vectors accessible to deep learning algorithms. Chapters on classical deep learning algorithms, like convolutional neural networks (CNN) and recurrent neural networks (RNN), demonstrate important NLP tasks as sentence classification and language generation. You will learn how to apply high-performance RNN models, like long short-term memory (LSTM) cells, to NLP tasks. You will also explore neural machine translation and implement a neural machine translator.
After reading this book, you will gain an understanding of NLP and you'll have the skills to apply TensorFlow in deep learning NLP applications, and how to perform specific NLP tasks.
What You Will Learn:
- Core concepts of NLP and various approaches to natural language processing
- How to solve NLP tasks by applying TensorFlow functions to create neural networks
- Strategies to process large amounts of data into word representations that can be used by deep learning applications
- Techniques for performing sentence classification and language generation using CNNs and RNNs
- About employing state-of-the art advanced RNNs, like long short-term memory, to solve complex text generation tasks
- How to write automatic translation programs and implement an actual neural machine translator from scratch
- The trends and innovations that are paving the future in NLP
Who this book is for:
This book is for Python developers with a strong interest in deep learning, who want to learn how to leverage TensorFlow to simplify NLP tasks. Fundamental Python skills are assumed, as well as some knowledge of machine learning and undergraduate-level calculus and linear algebra. No previous natural language processing experience required, although some background in NLP or computational linguistics will be helpful.
정보제공 :
목차
CONTENTS Preface = xi Chapter 1 : Introduction to Natural Language Processing = 1 What is Natural Language Processing? = 1 Tasks of Natural Language Processing = 2 The traditional approach to Natural Language Processing = 5 Understanding the traditional approach = 5 Drawbacks of the traditional approach = 10 The deep learning approach to Natural Language Processing = 10 History of deep learning = 11 The current state of deep learning and NLP = 13 Understanding a simple deep model - a Fully-Connected Neural Network = 14 The roadmap - beyond this chapter = 16 Introduction to the technical tools = 21 Description of the tools = 21 Installing Python and scikit-learn = 22 Installing Jupyter Notebook = 22 Installing TensorFiow = 23 Summary = 24 Chapter 2 : Understanding TensorFiow = 27 What is TensorFiow? = 28 Getting started with TensorFiow = 28 TensorFiow client in detail = 31 TensorFiow architecture - what happens when you execute the client? = 32 Cafe Le TensorFiow- understanding TensorFiow with an analogy = 35 Inputs, variables, outputs, and operations 36 Defining inputs in TensorFiow = 37 Defining variables in TensorFiow = 43 Defining TensorFiow oulputs = 45 Defining TensorFiow operations = 45 Reusing variables with scoping = 57 Implementing our first neural network = 59 Preparing the data = 60 Defining the TensorFiow graph = 61 Running the neural network = 63 Summary = 65 Chapter 3 : Word2vec - Learning Word Embeddings = 67 What is a word representation or meaning? = 69 Classical approaches to learning word representation = 69 Word Net - using an external lexical knowledge base for learning word representations = 70 One-hot encoded representation = 74 The TF-IDF method = 75 Co-occurrence matrix = 76 Word2vec- a neural network-based approach to learning word representation = 77 Exercise : is queen = king - he + she? = 78 Designing a loss function for learning word embeddings = 82 The skip-gram algorithm = 83 From raw text to structured data = 83 Learning the word embeddings with a neural network = 84 Implementing skip-gram with TensorFiow = 95 The Continuous Bag-of-Words algorithm = 98 Implementing CBOW in TensorFiow = 99 Summary = 100 Chapter 4 : Advanced Word2vec = 103 The original skip-gram algorithm = 104 Implementing the original skip-gram algorithm = 105 Comparing the original skip-gram with the improved skip-gram = 107 Comparing skip-gram with CBOW = 107 Performance comparison = 108 Which is the wirmer, skip-gram or CBOW? = 112 Extensions to the word embeddings algorithms = 114 Using the unigram distribution for negative sampling = 114 Implementing unigram-based negative sampling = 115 Subsampling - probabilistically ignoring the common words = 117 Implementing subsampling = 118 Comparing the CBOW and its extensions = 118 More recent algorithms extending skip-gram and CBOW = 119 A limitation of the skip-gram algorithm = 119 The structured skip-gram algorithm = 120 The loss function = 120 The continuous window model = 122 GloVe - Global Vectors representation = 123 Understanding GloVe = 123 Implementing GloVe = 125 Document classification with Word2vec = 126 Dataset = 127 Classifying documents with word embeddings = 127 Implementation -learning word embeddings = 128 Implementation- word embeddings to document embeddings = 129 Document clustering and t-SNE visualization of embedded documents = 130 Inspecting several outliers = 131 Implementation -clustering/classification of documents with K-means = 132 Summary = 134 Chapter 5 : Sentence Classification with Convolutional Neural Networks = 135 Introducing Convolution Neural Networks = 136 CNN fundamentals = 136 The power of Convolution Neural Networks = 139 Understanding Convolution Neural Networks = 139 Convolution operation = 140 Pooling operation = 144 Fully connected layers = 147 Putting everything together = 147 Exercise - image classification on MNIST with CNN = 148 About the data = 149 Implementing the CNN = 149 Analyzing the predictions produced with a CNN = 152 Using CNNs for sentence classification = 153 CNN structure = 153 Pooling over time = 157 Implementation - sentence classification with CNNs = 159 Summary = 162 Chapter 6 : Recurrent Neural Networks = 163 Understanding Recurrent Neural Networks = 164 The problem with feed-forward neural networks = 165 Modeling with Recurrent Neural Networks = 166 Technical description of a Recurrent Neural Network = 168 Backpropagation Through Time = 170 How backpropagation works = 170 Why we cannot use BP directly for RNNs = 171 Backpropagation Through Time - training RNNs = 172 Truncated BPTT- training RNNs efficiently = 173 Limitations of BPTT- vanishing and exploding gradients = 173 Applications of RNNs = 175 One-to-one RNNs = 176 One-to-many RNNs = 176 Many-to-one RNNs = 177 Many-to-many RNNs = 178 Generating text with RNNs = 179 Defining hyperparameters = 179 Unrolling the inputs over time for Truncated BPTT = 180 Defining the validation dataset = 181 Defining weights and biases = 181 Defining state persisting variables = 181 Calculating the hidden states and outputs with unrolled inputs = 182 Calculating the loss = 183 Resetting state at the beginning of a new segment of text = 183 Calculating validation output = 184 Calculating gradients and optimizing = 184 Outputting a freshly generated chunk of text = 184 Evaluating text results output from the RNN = 185 Perplexity- measuring the quality of the text result = 187 Recurrent Neural Networks with Context Features - RNNs with longer memory = 188 Technical description of the RNN-CF = 188 Implementing the RNN-CF = 190 Summary = 199 Chapter 7 : Long Short-Term Memory Networks = 201 Understanding Long Short-Term Memory Networks = 202 What is an LSTM? = 203 LSTMs in more detail = 204 How LSTMs differ from standard RNNs = 212 How LSTMs solve the vanishing gradient problem = 213 Improving LSTMs = 216 Greedy sampling = 217 Beam search = 218 Using word vectors = 219 Bidirectional LSTMs (BiLSTM) = 220 Other variants of LSTMs = 222 Peephole connections = 223 Gated Recurrent Units = 224 Summary = 226 Chapter 8 : Applications of LSTM - Generating Text = 229 Our data = 230 About the dataset = 230 Preprocessing data = 232 Implementing an LSTM = 232 Defining hyperparameters = 232 Defining parameters = 233 Defining an LSTM cell and its operations = 235 Defining inputs and labels = 236 Defining the optimizer = 238 Decaying learning rate over time = 238 Making predictions = 240 Calculating perplexity (loss) = 240 Resetting states = 240 Greedy sampling to break unimodality = 241 Generating new text = 241 Example generated text 242 Comparing LSTMs to LSTMs with peephole connections and GRUs = 243 Standard LSTM = 243 Gated Recurrent Units (GRUs) = 245 LSTMs with peepholes = 248 Training and validation perplexities over time = 250 Improving LSTMs - beam search = 251 Implementing beam search = 252 Examples generated with beam search = 254 Improving LSTMs -generating text with words instead of n-grams = 255 The curse of dimensionality = 255 Word2vec to the rescue = 255 Generating text with Word2vec = 256 Examples generated with LSTM-Word2vec and beam search = 258 Perplexity over time = 259 Using the TensorFiow RNN API = 260 Summary = 264 Chapter 9 : Applications of LSTM - Image Caption Generation = 265 Getting to know the data = 266 ILSVRC lmageNet dataset = 267 The MS-COCO dataset = 268 The machine learning pipeline for image caption generation = 269 Extracting image features with CNNs = 273 Implementation - loading weights and inferencing with VGG-16 = 274 Building and updating variables = 274 Preprocessing inputs = 275 Inferring VGG-16 = 277 Extracting vectorized representations of images = 278 Predicting class probabilities with VGG-16 = 278 Learning word embeddings = 280 Preparing captions for feeding into LSTMs = 281 Generating data for LSTMs = 282 Defining the LSTM = 284 Evaluating the results quantitatively = 287 BLEU = 287 ROUGE = 288 METEOR = 289 CIDEr = 291 BLEU-4 over time for our model = 292 Captions generated for test images = 293 Using TensorFiow RNN API with pretrained GloVe word vectors = 297 Loading GloVe word vectors = 298 Cleaning data = 299 Using pretrained embeddings with TensorFiow RNN API = 302 Summary = 308 Chapter 10 : Sequence-to-Sequence Learning - Neural Machine Translation = 311 Machine translation = 312 A brief historical tour of machine translation = 313 Rule-based translation = 313 Statistical Machine Translation (SMT) = 315 Neural Machine Translation (NMT) = 317 Understanding Neural Machine Translation = 320 Intuition behind NMT = 320 NMT architecture = 321 Preparing data for the NMT system = 325 At training time = 325 Reversing the source sentence = 326 At testing time = 327 Training the NMT = 328 Inference with NMT = 329 The BLEU score - evaluating the machine translation systems = 330 Modified precision = 331 Brevity penalty = 331 The final BLEU score = 332 Implementing an NMT from scratch - a German to English translator = 332 Introduction to data = 333 Preprocessing data = 333 Learning word embeddings = 335 Defining the encoder and the decoder = 335 Defining the end-to-end output calculation = 338 Some translation results = 340 Training an NMT jointly with word embeddings = 342 Maximizing matchings between the dataset vocabulary and the pretrained embeddings = 343 Defining the embeddings layer as a TensorFiow variable = 345 Improving NMTs = 348 Teacher forcing = 348 Deep LSTMs = 350 Attention = 351 Breaking the context vector bottleneck = 351 The attention mechanism in detail = 352 Some translation results- NMT with attention = 359 Visualizing attention for source and target sentences = 361 Other applications of Seq2Seq models - chatbots = 363 Training a chatbot = 364 Evaluating chatbots - Turing test = 365 Summary = 366 Chapter 11 : Current Trends and the Future of Natural Language Processing = 369 Current trends in NLP = 370 Word embeddings = 370 Neural Machine Translation (NMT) = 376 Penetration into other research fields = 378 Combining NLP with computer vision = 378 Reinforcement learning = 381 Generative Adversarial Networks for NLP = 384 Towards Artificial General Intelligence = 386 One Model to Learn Them All = 386 A joint many-task model - growing a neural network for multiple NLP tasks = 389 NLP for social media = 391 Detecting rumors in social media = 391 Detecting emotions in social media = 391 Analyzing political framing in tweets = 393 New tasks emerging = 393 Detecting sarcasm = 393 Language grounding = 394 Skimming text with LSTMs = 395 Newer machine learning models = 395 Phased LSTM = 396 Dilated Recurrent Neural Networks (DRNNs) = 397 Summary = 398 References = 398 Appendix : Mathematical Foundations and Advanced TensorFiow = 403 Basic data structures = 403 Scalar = 403 Vectors = 403 Matrices = 404 Indexing of a matrix = 405 Special types of matrices = 406 Identity matrix = 406 Diagonal matrix = 407 Tensors = 407 Tensor／matrix operations = 407 Transpose 407 Multiplication 408 Element-wise multiplication 409 Inverse 409 Finding the matrix inverse - SinJular Value Decomposition (SVD) 411 Norms 412 Determinant = 412 Probability = 413 Random variables = 413 Discrete random variables = 413 Continuous random variables = 414 The probability mass／density function = 414 Conditional probability = 417 Joint probability = 417 Marginal probability = 417 Bayes'''' rule = 418 Introduction to Keras = 418 Introduction to the TensorFiow seq2seq library = 421 Defining em beddings for the encoder and decoder = 421 Defining the encoder = 421 Defining the decoder = 422 Visualizing word embeddings with TensorBoard = 424 Starting TensorBoard = 424 Saving word embeddings and visualizing via TensorBoard = 425 Summary = 429 Other Books You May Enjoy = 431 Index = 437