What this book covers
Chapter 1, Introduction to Machine Learning, cover, the fundamentals of machine learning, its types, the machine learning life cycle, and applications of machine learning. We will also drill down into what it takes to become a certified TensorFlow developer.
Chapter 2, Introduction to TensorFlow, examines the TensorFlow ecosystem after which we set up our work environment. Also, we will take a look at data representation then we will build our hello world model using TensorFlow. We will conclude this chapter by examining how to debug and resolve error messages we come across as we build models.
Chapter 3, Linear Regression with TensorFlow, examines how to build linear regression models using TensorFlow. Next, we will explore various evaluation metrics for regression models. We will take matters a step further by building a salary prediction model with TensorFlow and we will close this chapter by mastering how to save and load models.
Chapter 4, Classification with TensorFlow, examines what classification modeling in machine learning is, and discusses different types of classification problems you may encounter in machine learning. Also, we’ll examine the various methods of evaluating classification problems and look at how to choose the right classification metrics for your use case. We will close this chapter by looking at a classification problem where we will learn how to build, compile, train, make predictions, and evaluate classification models using TensorFlow.
Chapter 5, Image Classification With Neural Networks, covers the anatomy of neural networks. We’ll discuss concepts such as forward propagation, backward propagation, and gradient descent. We will also discuss the moving parts such as the input layer, hidden layers, output layers, activation functions, loss function, and optimizers. We will close this chapter by examining how to build an image classifier using a neural network with TensorFlow.
Chapter 6, Improving the Model, examines various methods of improving the performance of our model. We will discuss the importance of data-centric strategies such as data augmentation and look at various hyperparameters, their impact, and how we can tune them to improve the performance of our models.
Chapter 7, Image Classification with Convolutional Neural Networks, introduces convolutional neural networks (CNNs). We’ll see how CNNs change the game in image classification tasks by exploring its anatomy, exploring concepts like convolutions and pooling operations. We will examine the challenges developers face when working with real world images. We will close the chapter by seeing CNNs in action as we apply them to classifying weather image data.
Chapter 8, Handling Overfitting, examines overfitting in greater details. We will examine what it is and why it occurs in real-world use cases. We will then go a step further by exploring various ways to overcome overfitting such as dropout regularization, early stopping and L1 & L2 regularization. We will put these ideas to the test using the weather image dataset case study to help us cement our understanding of these concepts.
Chapter 9, Transfer Learning, introduces the concept of transfer learning and we will discuss where and how we can apply transfer learning. We will also examine some best practices around applying transfer learning in our workflow. To close this chapter, we will build a real-world image classifier using pretrained models from TensorFlow.
Chapter 10, Introduction to Natural Language Processing, introduces natural language processing. We will discuss challenges around working with text data, and how we can move from language to vector representations. We will cover foundational ideas around text preprocessing and data preparation techniques such as tokenization, padding, sequencing, and word embedding. We will step things up by examining how to visualize word embeddings using TensorFlow’s projector. To close this chapter, we will build a sentiment classifier using TensorFlow.
Chapter 11, NLP with TensorFlow, goes deeper into the challenges of modeling text data. We will introduce recurrent neural networks (RNNs) and their variants, long short-term memory (LSTM) and gated recurrent units (GRU). We’ll see how they are uniquely tailored for handling sequential data such as text and time series data. We will apply these models in building a text classifier. We will also see how to apply pretrained word embeddings in this chapter. To close this chapter, we will see how we can build a children’s story generator using LSTMs.
Chapter 12, Introduction to Time Series, Sequences and Predictions, introduces time series data and examines the unique nature of time series data, its core characteristics, types and application of time series data. We will discuss some of the challenges with modeling time series data and examine a number of solutions. We will see how to prepare time series data for forecasting using utilities in TensorFlow, and then we will apply both statistical methods and machine learning techniques to forecast sales data for a fictional super store.
Chapter 13, Time Series, Sequences and Prediction with TensorFlow, discusses how to use in-built and custom learning rate schedulers in TensorFlow, we will also see how to apply lambda layers. We will see how to build time series forecasting models using RNNs, LSTMs, CNNs and CNN-LSTM networks. We will round off this chapter and the book by working through a problem where we will collect realworld closing stock prices and build a forecasting model.
