Machine learning isn’t as hard as it used to be. New books and courses come out daily, and it is becoming harder to pick the ones you’ll stick to. This list should help you get there somewhat quick, irrelevant of your background.
The list has something for everyone – no matter if you’re on a tight schedule or have little to none machine learning experience. The books aren’t ordered in any way, so pick the one you feel most comfortable with.
Let’s start with the first one.
The Hundred-Page Machine Learning Book
It’s more of a 150-page machine learning book, but you get the point. It’s designed for people in a hurry, as it doesn’t dwell that much on the details, but still manages to get everything important covered. It starts from the basics – introduction to machine learning and the most important statistical concepts (like random variables and Bayes theorem).
The story doesn’t end there, as every major supervised algorithm is well explained, from linear regression to support vector machines. Unsupervised algorithms are also covered, but later. Essential concepts from feature engineering, like one-hot encoding, normalization/standardization, and handling missing data are also covered.
The book also goes through the most important metrics for model evaluation, such as accuracy, precision, recall, and AUC/ROC curves. You can also expect to get some coverage of deep learning techniques, but the book is pretty vague there.
Overall, it is a no-brainer for anyone starting out or anyone who doesn’t have the time to sit through an 800-page book.
Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow
If you’re the type of person willing to spend months going through 800+ page book – congrats, you’re in for a treat. This book is a long-time best-seller on Amazon, just because it covers everything one might need to work in the field, explained with perfect clarity. Seriously, the book covers topics from the machine learning definition to GANs and reinforcement learning.
I’m glad this book was mandatory on my machine learning college course, as I’ve learned much more from it than from the professor. If your professor isn’t Andrew Ng, but more someone like Siraj, I reckon it will do the same.
As mentioned earlier, the books goes from simple topics, like data gathering, EDA, feature scaling, to actual machine learning through algorithms such as decision trees, random forest, and gradient boosting. It also covers the main dimensionality reduction techniques and unsupervised learning. All of that in the first 300 pages!
The rest is reserved for neural networks and deep learning, from theory to application in the TensorFlow library. Expect to learn a lot about ANNs, CNNs, RNNs, Autoencoders, GANs, and reinforcement learning.
Another no brainer if you have the time. And will.
Deep Learning for Coders with Fastai and PyTorch
PyTorch is my all-time favorite deep learning library, and FastAI is another one worth your time, as it uses PyTorch to develop production-ready models – you’ve guessed it – fast. It is aimed towards software developers, as it’s focused more on the practical part. It’s also perfectly suitable for data scientists not willing to dive too deep into the theory. The theoretical part is available later in the book, but I don’t see it as a mandatory read.
As mentioned before, the main focus is on the FastAI library, which is widely used to handle computer vision, tabular, NLP, and recommender system tasks with as little code as possible. The official website offers some free courses, so it’s a good idea to check these out.
The book covers the previously mentioned four topics in much depth, which is just enough to use the power of deep learning in your applications.
Overall, it’s an excellent book, definitely worth your time.
These are the three books I found particularly useful and easy enough to read. There’s no need to read all of them, so refer to the bullets below if you cannot decide:
- Read the first book if you’re starting with machine learning and don’t have much of free time
- Read the last book if you’re interested in practical machine learning and deep learning
- Read the second book if you have the time to learn both theoretical and practical aspects of the field
And that does it for this time. Thanks for reading.