Harsh Bhasin - Machine Learning for Beginners

My Mother

Harsh Bhasin is an Applied Machine Learning researcher. Mr. Bhasin worked as Assistant Professor in Jamia Hamdard, New Delhi, and taught as a guest faculty in various institutes including Delhi Technological University. Before that, he worked in C# Client-Side Development and Algorithm Development.

Mr. Bhasin has authored a few papers published in renowned journals including Soft Computing, Springer, BMC Medical Informatics and Decision Making, AI and Society, etc. He is the reviewer of prominent journals and has been the editor of a few special issues. He has been a recipient of a distinguished fellowship.

Outside work, he is deeply interested in Hindi Poetry, progressive era; Hindustani Classical Music, percussion instruments.

His areas of interest include Data Structures, Algorithms Analysis and Design, Theory of Computation , Python, Machine Learning and Deep learning.

Yogesh is the Chief Technology Officer at Byprice, a price comparison platform powered by advanced machine learning and deep learning models. He has successfully deployed 4 business critical applications in the last 2 years by harnessing the power of machine learning.

He has worked with recommendation systems, text similarity algorithms, deep learning models and image processing.

He is a visionary who understands how to drive product market fit for highly scalable solutions. He has 8 years of experience and has successfully deployed more than a dozen large scale B2B and B2C applications. He has worked as a senior software developer in one of Latin Americas largest e-commerce company, Linio, which serves 15 million users every month.

His vast experience in different fields of Software Engineering, Data Science and Storage Engines helps him in creating simple solutions for complex problems.

He graduated in Software Engineering from Delhi College of Engineering, INDIA.

He loves music, gardening and answering technical questions on StackOverflow.

YOU DONT HAVE TO BE GREAT TO START,
BUT YOU HAVE TO START TO BE GREAT.

ZIG ZIGLAR

I would like to thank a few people who helped me to start. Professor Moin Uddin, former Vice-Chancellor, Delhi Technological University has been a guiding light in my life. Late Professor A. K. Sharma had always encouraged me to do better and Professor Naresh Chauhan, YMCA Institute of Science and Technology, Faridabad has always been supportive.

I would also like to thank my students Aayush Arora, Arush Jasuja, and Deepanshu Goel for their help. I would also like to thank BPB Publications for giving all the support provided when needed. Also would like to thank Yogesh for his efforts, for the feedback given by him.

Lastly, I would like to thank my mother and sister, my friends, and my pets: Zoe and Xena for bearing me.

Data is being collected by websites, mobile applications, dispensations (on various pretexts), and even by devices. This data must be analyzed to become useful. The patterns extracted by this data can be used for targeted marketing, for national security, for propagating believes and myths, and for many other tasks. Machine Learning helps us in explaining the data by a simple model. It is currently being used in various disciplines ranging from Biology to Finance and hence has become one of the most important subjects.

There is an immediate need for a book that not only explains the basics but also includes implementations. The analysis of the models using various datasets needs to be explained, to find out which model can be used to explain a given data. Despite the presence of excellent books on the subject, none of the existing books covers all the above points.

This book covers major topics in Machine Learning. It begins with data cleansing and presents a brief overview of visualization. The first chapter of this book talks about introduction to Machine Learning, training and testing, cross-validation, and feature selection. The second chapter presents the algorithms and implementation of the most common feature selection techniques like Fisher Discriminant ratio and mutual information.

The third chapter introduces readers to Linear Regression and Gradient Descent. The later would be used by many algorithms that would be discussed later in the book. Some of the important classification techniques like K-nearest neighbors, logistic regression, Nave Bayesian, and Linear Discriminant Analysis have been discussed and implemented in the next chapter. The next two chapters focus on Neural Networks and their implementation. The chapters systematically explain the biological background, the limitations of the perceptron, and the backpropagation model. The Support Vector Machines and Kernel methods have been discussed in the next chapter. This is followed by a brief overview and implementation of Decision Trees and Random Forests.

Various feature extraction techniques have been discussed in the book. These include Fourier Transform, STFT, and Local Binary patterns. The book also discusses Principle Component Analysis and its implementation.