Chapter 1, Machine Learning Review, is a refresher of basic concepts and techniques that the reader would have learned from Packt's Learning Machine Learning in Java or a similar text. This chapter is a review of concepts such as data, data transformation, sampling and bias, features and their importance, supervised learning, unsupervised learning, big data learning, stream and real-time learning, probabilistic graphic models, and semi-supervised learning.

Chapter 2, Practical Approach to Real-World Supervised Learning, cobwebs dusted, dives straight into the vast field of supervised learning and the full spectrum of associated techniques. We cover the topics of feature selection and reduction, linear modeling, logistic models, non-linear models, SVM and kernels, ensemble learning techniques such as bagging and boosting, validation techniques and evaluation metrics, and model selection. Using WEKA and RapidMiner, we carry out a detailed case study, going through all the steps from data analysis to analysis of model performance. As in each of the other chapters, the case study is presented as an example to help the reader understand how the techniques introduced in the chapter are applied in real life. The dataset used in the case study is UCI HorseColic.

Chapter 3, Unsupervised Machine Learning Techniques, presents many advanced methods in clustering and outlier techniques, with applications. Topics covered are feature selection and reduction in unsupervised data, clustering algorithms, evaluation methods in clustering, and anomaly detection using statistical, distance, and distribution techniques. At the end of the chapter, we perform a case study for both clustering and outlier detection using a real-world image dataset, MNIST. We use the Smile API to do feature reduction and ELKI for learning.

Chapter 4, Semi-supervised Learning and Active Learning, gives details of algorithms and techniques for learning when only a small amount labeled data is present. Topics covered are self-training, generative models, transductive SVMs, co-training, active learning, and multi-view learning. The case study involves both learning systems and is performed on the real-world UCI Breast Cancer Wisconsin dataset. The tools introduced are JKernelMachines ,KEEL and JCLAL.

Chapter 5, Real-Time Stream Machine Learning, covers data streams in real-time present unique circumstances for the problem of learning from data. This chapter broadly covers the need for stream machine learning and applications, supervised stream learning, unsupervised cluster stream learning, unsupervised outlier learning, evaluation techniques in stream learning, and metrics used for evaluation. A detailed case study is given at the end of the chapter to illustrate the use of the MOA framework. The dataset used is Electricity (ELEC).

Chapter 6, Probabilistic Graph Modeling, shows that many real-world problems can be effectively represented by encoding complex joint probability distributions over multi-dimensional spaces. Probabilistic graph models provide a framework to represent, draw inferences, and learn effectively in such situations. The chapter broadly covers probability concepts, PGMs, Bayesian networks, Markov networks, Graph Structure Learning, Hidden Markov Models, and Inferencing. A detailed case study on a real-world dataset is performed at the end of the chapter. The tools used in this case study are OpenMarkov and WEKA's Bayes network. The dataset is UCI Adult (Census Income).

Chapter 7, Deep Learning, If there is one super-star of machine learning in the popular imagination today it is deep learning, which has attained a dominance among techniques used to solve the most complex AI problems. Topics broadly covered are neural networks, issues in neural networks, deep belief networks, restricted Boltzman machines, convolutional networks, long short-term memory units, denoising autoencoders, recurrent networks, and others. We present a detailed case study showing how to implement deep learning networks, tuning the parameters and performing learning. We use DeepLearning4J with the MNIST image dataset.

Chapter 8, Text Mining and Natural Language Processing, details the techniques, algorithms, and tools for performing various analyses in the field of text mining. Topics broadly covered are areas of text mining, components needed for text mining, representation of text data, dimensionality reduction techniques, topic modeling, text clustering, named entity recognition, and deep learning. The case study uses real-world unstructured text data (the Reuters-21578 dataset) highlighting topic modeling and text classification; the tools used are MALLET and KNIME.

Chapter 9, Big Data Machine Learning – the Final Frontier, discusses some of the most important challenges of today. What learning options are available when data is either big or available at a very high velocity? How is scalability handled? Topics covered are big data cluster deployment frameworks, big data storage options, batch data processing, batch data machine learning, real-time machine learning frameworks, and real-time stream learning. In the detailed case study for both big data batch and real-time we select the UCI Covertype dataset and the machine learning libraries H2O, Spark MLLib and SAMOA.

Appendix A, Linear Algebra, covers concepts from linear algebra, and is meant as a brief refresher. It is by no means complete in its coverage, but contains a whirlwind tour of some important concepts relevant to the machine learning techniques featured in the book. It includes vectors, matrices and basic matrix operations and properties, linear transformations, matrix inverse, eigen decomposition, positive definite matrix, and singular value decomposition.

Appendix B, Probability, provides a brief primer on probability. It includes the axioms of probability, Bayes' theorem, density estimation, mean, variance, standard deviation, Gaussian standard deviation, covariance, correlation coefficient, binomial distribution, Poisson distribution, Gaussian distribution, central limit theorem, and error propagation.

This book assumes you have some experience of programming in Java and a basic understanding of machine learning concepts. If that doesn't apply to you, but you are curious nonetheless and self-motivated, fret not, and read on! For those who do have some background, it means that you are familiar with simple statistical analysis of data and concepts involved in supervised and unsupervised learning. Those who may not have the requisite math or must poke the far reaches of their memory to shake loose the odd formula or funny symbol, do not be disheartened. If you are the sort that loves a challenge, the short primer in the appendices may be all you need to kick-start your engines—a bit of tenacity will see you through the rest! For those who have never been introduced to machine learning, the first chapter was equally written for you as for those needing a refresher—it is your starter-kit to jump in feet first and find out what it's all about. You can augment your basics with any number of online resources. Finally, for those innocent of Java, here's a secret: many of the tools featured in the book have powerful GUIs. Some include wizard-like interfaces, making them quite easy to use, and do not require any knowledge of Java. So if you are new to Java, just skip the examples that need coding and learn to use the GUI-based tools instead!

The primary audience of this book is professionals who works with data and whose responsibilities may include data analysis, data visualization or transformation, the training, validation, testing and evaluation of machine learning models—presumably to perform predictive, descriptive or prescriptive analytics using Java or Java-based tools. The choice of Java may imply a personal preference and therefore some prior experience programming in Java. On the other hand, perhaps circumstances in the work environment or company policies limit the use of third-party tools to only those written in Java and a few others. In the second case, the prospective reader may have no programming experience in Java. This book is aimed at this reader just as squarely as it is at their colleague, the Java expert (who came up with the policy in the first place).

A secondary audience can be defined by a profile with two attributes alone: an intellectual curiosity about machine learning and the desire for a single comprehensive treatment of the concepts, the practical techniques, and the tools. A specimen of this type of reader can opt to skip the math and the tools and focus on learning the most common supervised and unsupervised learning algorithms alone. Another might skim over Chapters 1, 2, 3, and 7, skip the others entirely, and jump headlong into the tools—a perfectly reasonable strategy if you want to quickly make yourself useful analyzing that dataset the client said would be here any day now. Importantly, too, with some practice reproducing the experiments from the book, it'll get you asking the right questions of the gurus! Alternatively, you might want to use this book as a reference to quickly look up the details of the algorithm for affinity propagation (Chapter 3, Unsupervised Machine Learning Techniques), or remind yourself of an LSTM architecture with a brief review of the schematic (Chapter 7, Deep Learning), or dog-ear the page with the list of pros and cons of distance-based clustering methods for outlier detection in stream-based learning (Chapter 5, Real-Time Stream Machine Learning). All specimens are welcome and each will find plenty to sink their teeth into.

In this book, you will find a number of text styles that distinguish between different kinds of information. Here are some examples of these styles and an explanation of their meaning.

Code words in text, database table names, folder names, filenames, file extensions, pathnames, dummy URLs, user input, and Twitter handles are shown as follows: "The algorithm calls the eliminate function in a loop, as shown here."

A block of code is set as follows:

DataSource source = new DataSource(trainingFile);
Instances data = source.getDataSet();
if (data.classIndex() == -1)
  data.setClassIndex(data.numAttributes() - 1);

Any command-line input or output is written as follows:

Correctly Classified Instances     53       77.9412 %
Incorrectly Classified Instances    15       22.0588 %

New terms and important words are shown in bold.

Feedback from our readers is always welcome. Let us know what you think about this book—what you liked or disliked. Reader feedback is important for us as it helps us develop titles that you will really get the most out of.

To send us general feedback, simply e-mail <[email protected]>, and mention the book's title in the subject of your message.

If there is a topic that you have expertise in and you are interested in either writing or contributing to a book, see our author guide at www.packtpub.com/authors.

Now that you are the proud owner of a Packt book, we have a number of things to help you to get the most from your purchase.

You can download the code files by following these steps:

You can also download the code files by clicking on the Code Files button on the book's webpage at the Packt Publishing website. This page can be accessed by entering the book's name in the Search box. Please note that you need to be logged in to your Packt account.

Once the file is downloaded, please make sure that you unzip or extract the folder using the latest version of:

The code bundle for the book is also hosted on GitHub at https://github.com/mjmlbook/mastering-java-machine-learning. We also have other code bundles from our rich catalog of books and videos available at https://github.com/PacktPublishing/. Check them out!