Chapter 1, The Realm of Supervised Learning, covers various machine learning paradigms that will help you to understand how the field is divided into multiple subgroups. This chapter briefly discuss the differences between supervised and unsupervised learning, along with the concepts of regression, classification, and clustering. We will learn how to preprocess data for machine learning. We will discuss regression analysis in detail and learn how to apply it to a couple of real-world problems, including house price estimation and bicycle demand distribution.
Chapter 2, Constructing a Classifier, shows you how to perform data classification using various models. We will discuss techniques including logistic regression and the naïve Bayes model. We will learn how to evaluate the accuracy of classification algorithms. We will discuss the concept of cross-validation and learn how to use it to validate our machine learning model. We will learn about validation curves and how to plot them. We will apply these supervised learning techniques to real-world problems, such as income bracket estimation and activity recognition.
Chapter 3, Predictive Modeling, covers the premise of predictive modeling and why it’s needed. We will learn about SVMs and understand how they work. We will learn how to use them to classify data. We will discuss the concept of hyperparameters and how they affect the performance of SVMs. We will learn how to use grid search to find the optimal set of hyperparameters. We will discuss how to estimate the confidence measure of the outputs. We will talk about ensemble learning and the various algorithms in this group, such as decision trees and random forests. We will then learn how to apply these techniques to real-world event prediction.
Chapter 4, Clustering with Unsupervised Learning, covers the concept of unsupervised learning and what we hope to achieve from it. We will learn how to perform data clustering and how to apply the k-means clustering algorithm to do it. We will visualize the clustering process using sample data. We will discuss mixture models and Gaussian mixture models. We will then apply these techniques to perform market segmentation using customer information.
Chapter 5, Visualizing Data, discusses how to visualize data and explains why it's useful for machine learning. We will learn how to use Matplotlib to interact with our data and visualize it using various techniques. We will discuss histograms and how they are useful. We will explore different methods for visualizing data, including line charts, scatter plots, and bubble plots. We will learn how to use heat maps, perform animation, and do 3D plotting.
Chapter 6, Building Recommendation Engines, introduces recommendation engines and shows us how to use it for checking movie recommendations. We will construct a k-nearest neighbors classifier to find similar users in our dataset and then generate movie recommendations using a filtering model with TensorFlow.
Chapter 7, Analyzing Text Data, shows you how to analyze text data. We will understand various concepts such as the bag-of-words model, tokenization, and stemming. We will learn about the features that can be extracted from text. We will discuss how to build a text classifier. We will then use these techniques to infer the sentiment of a sentence. We will also learn how to automatically identify the topic of an unknown paragraph. We will then move on to evaluating regression and classification models, and then step into recipes that can help us with selecting models.
Chapter 8, Speech Recognition, demonstrates how you can work with speech data. We will learn about concepts including windowing and convolution. We will understand how to extract features from speech data. We will learn about hidden Markov models and how to use them to automatically recognize the words being spoken.
Chapter 9, Dissecting Time Series and Sequential Data, introduces the concept of structured learning. We will come to understand the various characteristics of time series data. We will learn about conditional random fields and see how to use them for prediction. We will then use this technique to analyze stock market data.
Chapter 10, Image Content Analysis, shows how to analyze images. We will learn how to detect keypoints and extract features from images. We will discuss the concept of a bag of visual words and see how it applies to image classification. We will learn how to build a visual code book and extract feature vectors for image classification. We will then understand how to use extremely random forests to perform object recognition.
Chapter 11, Biometric Face Recognition, shows how to perform face recognition. We will understand the differences between face detection and face recognition. We will talk about dimensionality reduction and how to use PCA to achieve this. We will learn about Fisher Faces and how it can be used for face recognition. We will perform face detection on a live video. We will then use these techniques to identify the person in front of the camera.
Chapter 12, Reinforcement Learning Techniques, discusses reinforcement learning techniques and its applications. It also discusses the elements of reinforcement learning setup, approaches to reinforcement learning, and its challenges, along with topics such as Markov decision processes, the exploration-exploitation dilemma, discounted future reward, and Q-learning.
Chapter 13, Deep Neural Networks, discusses deep neural networks. We will learn about perceptrons and see how they are used to build neural networks. We will explore the interconnections between multiple layers in a deep neural network. We will discuss how a neural network learns about the training data and builds a model. We will learn about the cost function and backpropagation. We will then use these techniques to perform optical character recognition. We will work with different frameworks for deep learning, including TensorFlow, PyTorch, and Caffe.
Chapter 14, Unsupervised Representation Learning, discusses the problem of learning representations for data such as images, videos, and natural language corpuses in an unsupervised manner. We will go through autoencoders and their applications, word embeddings, and t-SNEs. We will also use denoising autoencoders to detect fraudulent transactions using word embeddings. Lastly, we will move on to implementing LDA with the help of various recipes.
Chapter 15, Automated Machine Learning and Transfer Learning, discusses recipes based on automated machine learning and transfer learning. We will learn how to work with Auto-WEKA and how to use AutoML to generate machine learning pipelines. We will learn how to work with Auto-Keras and then move on to using MLBox for leak detection. Furthermore, we will learn how to implement transfer learning with the help of multiple recipes.
Chapter 16, Unlocking Production Issues, discusses production-related issues. We will go through how a reader can handle unstructured data, along with how we can keep a track of changes in our machine learning models. We will also learn how to optimize a retraining schedule and how to deploy our machine learning models.