TensorFlow Machine Learning Cookbook

Over 60 recipes to build intelligent machine learning systems with the power of Python, 2nd Edition

Nick McClure

• Editorial: Packt Publishing
• Año: 2018
• ISBN:
  9781789130768
• Edición: 2
• Categoría: Informatique
• Categoría: Intelligence artificielle (IA) et sémantique

Skip the theory and get the most out of Tensorflow to build production-ready machine learning models

Key Features

• Exploit the features of Tensorflow to build and deploy machine learning models
• Train neural networks to tackle real-world problems in Computer Vision and NLP
• Handy techniques to write production-ready code for your Tensorflow models

Book Description

TensorFlow is an open source software library for Machine Intelligence. The independent recipes in this book will teach you how to use TensorFlow for complex data computations and allow you to dig deeper and gain more insights into your data than ever before.

With the help of this book, you will work with recipes for training models, model evaluation, sentiment analysis, regression analysis, clustering analysis, artificial neural networks, and more. You will explore RNNs, CNNs, GANs, reinforcement learning, and capsule networks, each using Google's machine learning library, TensorFlow. Through real-world examples, you will get hands-on experience with linear regression techniques with TensorFlow. Once you are familiar and comfortable with the TensorFlow ecosystem, you will be shown how to take it to production.

By the end of the book, you will be proficient in the field of machine intelligence using TensorFlow. You will also have good insight into deep learning and be capable of implementing machine learning algorithms in real-world scenarios.

What you will learn

• Become familiar with the basic features of the TensorFlow library
• Get to know Linear Regression techniques with TensorFlow
• Learn SVMs with hands-on recipes
• Implement neural networks to improve predictive modeling
• Apply NLP and sentiment analysis to your data
• Master CNN and RNN through practical recipes
• Implement the gradient boosted random forest to predict housing prices
• Take TensorFlow into production

Who this book is for

If you are a data scientist or a machine learning engineer with some knowledge of linear algebra, statistics, and machine learning, this book is for you. If you want to skip the theory and build production-ready machine learning models using Tensorflow without reading pages and pages of material, this book is for you. Some background in Python programming is assumed.

Recurrent Neural Networks

In this chapter, we will cover recurrent neural networks (RNNs) and how to implement them in TensorFlow. We will start by demonstrating how to use an RNN to predict spam. We will then introduce a variant on RNNs for creating Shakespearean text. We will finish by creating an RNN sequence-to-sequence model to translate from English to German:

• Implementing RNNs for spam prediction
• Implementing an LSTM model
• Stacking multiple LSTM layers
• Creating sequence-to-sequence models
• Training a Siamese similarity measure

All the code to this chapter can be found online at https://github.com/nfmcclure/tensorflow_cookbook and at the Packt online repository: https://github.com/PacktPublishing/TensorFlow-Machine-Learning-Cookbook-Second-Edition.

Introduction

Of all the machine learning algorithms we have considered thus far, none have considered data as a sequence. To take sequence data into account, we extend neural networks that store outputs from prior iterations. This type of neural network is called an RNN. Consider the fully connected network formulation:

Here, the weights are given by A multiplied by the input layer, x, and then run through an activation function,

, which gives the output layer, y.

If we have a sequence of input data,

, we can adapt the fully connected layer to take prior inputs into account, as follows:

On top of this recurrent iteration to get the next input, we want to get the probability distribution output, as follows:

Once we have a full sequence output,

, we can consider the target as a number or category by just considering the last output. See the following diagram for how a general architecture might work:

Figure 1: To predict a single number, or a category, we take a sequence of inputs (tokens) and consider the final output as the predicted output

We can also consider the sequence output as an input in a sequence-to-sequence model:

Figure 2: To predict a sequence, we may also feed the outputs back into the model to generate multiple outputs

For arbitrarily long sequences, training with the back-propagation algorithm creates long time-dependent gradients. Because of this, there exists a vanishing or exploding gradient problem. Later in this chapter, we will explore a solution to this problem by expanding the RNN cell into what is called a Long Short Term Memory (LSTM) cell. The main idea is that the LSTM cell introduces another operation, called a gate, which controls the flow of information through the sequence. We will go over the details in later chapters.

When dealing with RNN models for NLP, encoding is a term used to describe the process of converting data (words or characters in NLP) into numerical RNN features. The term decoding is the process of converting RNN numerical features into output words or characters.

Implementing RNN for spam prediction

To start, we will apply the standard RNN unit to predict a singular numerical output, a probability of being spam.

Getting ready

In this recipe, we will implement a standard RNN in TensorFlow to predict whether or not a text message is spam or ham. We will use the SMS spam-collection dataset from the ML repository at UCI. The architecture we will use for prediction will be an input RNN sequence from embedded text, and we will take the last RNN output as a prediction of spam or ham (1 or 0).

How to do it...

1. We start by loading the libraries required for this script:

import os import re import io import requests import numpy as np import matplotlib.pyplot as plt import tensorflow as tf from zipfile import ZipFile 

2. Next, we start a graph session and set the RNN model parameters. We will run the data through 20 epochs, in batch sizes of 250. The maximum length of each text we will consider is 25 words; we will cut longer texts to 25 or zero-pad shorter texts. The RNN will be size 10 units. We will only consider words that appear at least 10 times in our vocabulary, and every word will be embedded in a trainable vector of size 50. The dropout rate will be a placeholder that we can set at 0.5 during training time or 1.0 during evaluation:

sess = tf.Session() epochs = 20 batch_size = 250 max_sequence_length = 25 rnn_size = 10 embedding_size = 50 min_word_frequency = 10 learning_rate = 0.0005 dropout_keep_prob = tf.placeholder(tf.float32) 

3. Now we get the SMS text data. First, we check whether it has already been downloaded and, if so, read in the file. Otherwise, we download the data and save it:

data_dir = 'temp' data_file = 'text_data.txt' if not os.path.exists(data_dir): os.makedirs(data_dir) if not os.path.isfile(os.path.join(data_dir, data_file)): zip_url = 'http://archive.ics.uci.edu/ml/machine-learning-databases/00228/smsspamcollection.zip' r = requests.get(zip_url) z = ZipFile(io.BytesIO(r.content)) file = z.read('SMSSpamCollection') # Format Data text_data = file.decode() text_data = text_data.encode('ascii',errors='ignore') text_data = text_data.decode().split('\n') # Save data to text file with open(os.path.join(data_dir, data_file), 'w') as file_conn: for text in text_data: file_conn.write("{}\n".format(text)) else: # Open data from text file text_data = [] with open(os.path.join(data_dir, data_file), 'r') as file_conn: for row in file_conn: text_data.append(row) text_data = text_data[:-1] text_data = [x.split('\t') for x in text_data if len(x)>=1] [text_data_target, text_data_train] = [list(x) for x in zip(*text_data)] 

4. To reduce our vocabulary, we will clean the input text by removing special characters, and extra white space, and putting everything in lowercase:

def clean_text(text_string):
 text_string = re.sub(r'([^sw]|_|[0-9])+', '', text_string)
 text_string = " ".join(text_string.split())
 text_string = text_string.lower()
 return text_string

# Clean texts
text_data_train = [clean_text(x) for x in text_data_train]

Note that our cleaning step removes special characters. As an alternative, we could also have replaced them with a space. Ideally, this depends on the formatting of the dataset.

5. Now we process the text with a built-in vocabulary processor function from TensorFlow. This will convert the text to an appropriate list of indices:

vocab_processor = tf.contrib.learn.preprocessing.VocabularyProcessor(max_sequence_length, min_frequency=min_word_frequency) text_processed = np.array(list(vocab_processor.fit_transform(text_data_train))) 

Note that the functions in contrib.learn.preprocessing are currently deprecated, (using the current TensorFlow version, 1.10). The current replacement suggestion, the TensorFlow-preprocessing package, only runs in Python 2 as of now. The effort to move TensorFlow-preprocessing to Python 3 is currently underway and will replace the previous two lines. Remember that all current and up-to-date code can be found on this GitHub page: https://www.github.com/nfmcclure/tensorflow_cookbook and at the Packt repository: https://github.com/PacktPublishing/TensorFlow-Machine-Learning-Cookbook-Second-Edition.

6. Next, we shuffle the data to randomize it:

text_processed = np.array(text_processed) text_data_target = np.array([1 if x=='ham' else 0 for x in text_data_target]) shuffled_ix = np.random.permutation(np.arange(len(text_data_target))) x_shuffled = text_processed[shuffled_ix] y_shuffled = text_data_target[shuffled_ix] 

7. We also split the data into an 80-20 train-test dataset:

ix_cutoff = int(len(y_shuffled)*0.80) x_train, x_test = x_shuffled[:ix_cutoff], x_shuffled[ix_cutoff:] y_train, y_test = y_shuffled[:ix_cutoff], y_shuffled[ix_cutoff:] vocab_size = len(vocab_processor.vocabulary_) print("Vocabulary Size: {:d}".format(vocab_size)) print("80-20 Train Test split: {:d} -- {:d}".format(len(y_train), len(y_test)))

For this recipe, we will not be doing any hyperparameter tuning. If the reader goes in this direction, remember to split up the dataset into train-test-validation sets before proceedin...