Machine Learning Algorithms By Giuseppe Bonaccorso

machine learning algorithm book review

Machine Learning Algorithms (Packt Publishing) is a book written for all computer scientists who desire to enter the world of machine learning starting a progressive path starting from the fundamental elements and arriving at the basics of Deep Learning.

The book covers both the mathematical theory and the practical aspects with a continuous reference to Scikit-Learn and other Python libraries. In this way, the reader can immediately test what he has learned, rerunning the examples or modifying them in order to solve different problems.

The main topics are focused both on supervised and unsupervised learning, in particular: Linear and Logistic Regression, SVM, Neural Networks, Decision Trees, Random Forests and Ensemble Learning. A chapter is dedicated to the data pre-processing, with a focus on linear and non-linear dimensionality reduction methods like PCA, NMF and Kernel-PCA.

Special chapters are dedicated to the Recommendation Systems based on different algorithms, with a focus on the Collaborative Filtering and a parallelizable Spark implementation.

Two chapters are dedicated to the basics of Natural Language Processing, with examples of tokenization, stemming, lemmatization and vectorization based on NLTK, Sentiment Analysis and Topic Modeling, including examples of Latent Semantic Analysis and Latent Dirichlet Allocation.

The last chapters are dedicated to the introduction of Deep Learning, with brief introductions of Tensorflow and Keras and example of gradient computations, neural networks and convolution neural networks for image classification.

The book is clear and avoids mathematical complications when not necessary. This pragmatic approach allows computer scientists with a basic Python knowledge to focus their efforts in designing the solutions without worrying about the mathematical complexity of many algorithms. At the same time, the rationale of all strategies is always explained, with references to specific papers and books that can be studied to have a deeper understanding.

All provided source code is in Python and it’s based on Scikit-Learn, Matplotlib, Scipy, Numpy, Crab, Pyspark, NLTK, Tensorflow and Keras.

To learn more about the author, please visit this link.

You can also buy the book here on Amazon.

I Write Things.