Different Ways To Master Quantum Machine Learning
I learned Quantum Machine Learning the hard way. But there’s a better way
This is the hard way
I did not have the fortune to take a quantum computing class in college. Not to speak of a class in quantum machine learning. At the time, it wouldn’t have been much fun anyway. In the early 2000s, quantum computing was just about to take the step from a pure theory to be evaluated in research labs. It was a field for theoretical physicists and mathematicians.
At the time, I haven’t even heard about it. When I heard about quantum computing for the first time, I think it was around 2008, researchers had successfully entangled qubits and were able to control them. Of course, Star Trek-like transportation came to mind when I heard two particles that were physically apart could share a state so that it was possible to change the state of one particle by observing the other.
Yet, until around 2014, I did not pay much attention. I was too busy writing my doctoral dissertation about assessing the effort caused by the requirements in a software development project. When I returned to normal life, I was just right in time to experience the end of the second AI winter and the advent of practical machine learning. What had been theory thus far became reality now.
When I got into machine learning, the field was already quite evolved. Libraries such as Scikit-Learn, later Keras, TensorFlow, and PyTorch made the development of machine learning algorithms convenient. Even though my favorite books were published sometime later, there were already a lot of good books and learning material available.
Note: My favorite books are: Hands-On Machine Learning with Scikit-Learn and TensorFlow by Aurélien Géron released 2017 and Deep Learning with Python by Francois Chollet released 2018.
But the models we’re developing today become increasingly hard to train. Open AI’s GPT-3 model that uses deep learning to produce human-like text would require 355 years on a single GPU and cost $4,600,000 to train. It is hard to believe that the upcoming milestones can be reached classically.
This insight brought quantum computing back into my focus. Quantum computing promises to reduce the computational complexity of certain algorithms by magnitudes. It promises to solve tasks in a few seconds classical computers would need thousands of years for. It promises to prevent us from the next AI winter that would be caused by the inability to reach the next milestones of machine learning.
In 2018, I started to deep dive into quantum machine learning. Scientific papers and a few theoretical books were all I could find. And these did not cover quantum machine learning but quantum computing in general. I was happy about every little piece.
After reading these quantum computing publications, I was left scratching my head. Most of the papers are pretty heavy on math and assume you’re familiar with a lot of physical jargon. I could not even find an appropriate starting point or some guidance on how to structure my learning efforts.
Frustrated with my failed attempts, I spent hours searching on Google. I hunted for quantum tutorials, only to come up empty-handed.
I could clearly see the potential value of quantum computing for machine learning. Yet, I couldn’t see how all these parts of quantum computing fit together. Entry-level material was hard to find. And practical guides were simply not existent. I wanted to get started, but I had nothing to show for my effort, except for a stack of quantum computing papers on my desk that I hardly understood.
Finally, I resorted to learning the theory first. I heard about Qiskit, the IBM quantum SDK for Python. At the time, its documentation was rather poor, especially if you were not familiar with all the physical jargon and its underlying theory. But it let me experience what some of these things like superposition, entanglement, and interference meant in practice.
This practical knowledge enabled me to connect quantum computing with the algorithms I knew from machine learning. I found my way to quantum machine learning success through myriads of trial-and-error experiments, countless late nights, and a lot of endurance.
I really believe that painstakingly working everything out in small pieces made an impact on how I understand quantum machine learning. Though, I would recommend not taking the same path.
This is the better way
My personal takeaways are:
You don’t need to cram all the theory before you start applying it
You don’t need to work through tons of equations
You don’t need to be a mathematician to master quantum machine learning
You don’t need to be a physicist to understand quantum machine learning
You’ll do great as a programmer, an engineer, a data scientist, or any other profession.
But quantum machine learning is taught the wrong way
When I started studying the quantum part of quantum machine learning, I took a deep dive into the theory and into math. Because this is what most quantum computing resources focus on.
Of course, it is desirable to have an understanding of the underlying math and the theory. But more importantly, you need to have an understanding of what the concepts mean in practice. If you know what you can do and how you need to do it, you don’t need to think about how it (physically) works all the time.
Don’t get me wrong. In quantum machine learning, physics and math are important. But if you don’t use the theoretical knowledge and apply it to solve real-world tasks, then you’ll have a hard time finding your space in the quantum machine learning world. You need to become a quantum machine learning practitioner from the very beginning.
In contrast to the days when I started, today there are quite a few resources available. But most of them fall into one of the following categories
Theoretical papers with lots of equations prove some quantum speedup of an algorithm. Yet, they don’t show any code.
Textbooks on quantum computing in general explain the concepts in an understandable manner. But they are short on showing how to use them for a purpose.
Blog posts show you an actual algorithm in code. But they don’t relate the code to any underlying concept. You see it works. But you don’t learn anything about why and how it works.
By no means do I want to say these resources are not worth reading. But these resources are not useful to learn how to apply quantum machine learning. For someone just about to start with quantum machine learning, you would need to invest a lot of time and effort for little to no practical return.
There is a fundamental disconnect between theory and practice. There’s a gap I want to help to fill with Hands-On Quantum Machine Learning with Python so you can learn in a more efficient — a better way.
This is the book I wish I had when I first started studying quantum machine learning. Inside this book, you’ll find practical walkthroughs and hands-on tutorials with lots of code. The book introduces new theory just in time you need it to take the next step. You’ll learn a lot of theory. But you’re not left alone with it. We directly apply our newly acquired knowledge to solve an actual problem.
We will not only implement different quantum machine learning algorithms, such as Quantum Naïve Bayes and Quantum Bayesian Networks. But we also use them to solve actual problems taken from Kaggle.
By the time you finish this book, you’ll know these algorithms, what they do, why you need them, how they work, and most importantly how to use them.
Hands-On Quantum Machine Learning With Python strives to be the perfect balance between theory taught in a textbook and the actual hands-on knowledge you’ll need to implement real-world solutions.
This book is your comprehensive guide to get started with Quantum Machine Learning”–the use of quantum computing for machine learning tasks.
Don’t be late for the party!
Get your copy by supporting the Kickstarter Campaign
