The world’s leading publication for data science, AI, and ML professionals.

Quantum Machine Learning – It’s time to start now

QML the next logical step for ML practitioner

Image by Gerd Altmann from Pixabay
Image by Gerd Altmann from Pixabay

In the coming decades, quantum computation will play an essential role in technology, science and business advancement. With Google claiming "Quantum supremacy" and exponential interest in this subject, we aren’t far from seeing many intractable issues challenging to tackle with a classical computer solved by Quantum processing. Furthermore, there is a clear indication that Machine Learning and quantum computation will play complementary roles in strengthening each other’s areas.

But the main question is why now one should dive into a complex topic like Quantum Machine Learning that looks complex and difficult to learn. Fortunately, Quantum Computation and then Quantum Machine learning is not so scary as it appears. Thus, one can jump-start without worrying too much about the complexity of many "spooky" functions of Quantum Mechanics that make one frightened.

Why should one learn Quantum Machine Learning?

Below are a few reasons why you should seriously consider diving into Quantum Computation(QC) and then into Quantum Machine Learning (QML).

Image by Author
Image by Author
  1. Math is extremely elegant, and it’s fun to understand Quantum computing using Mathematics.
  2. Excellent free resources and reference material like books are available for learning.
  3. The learning curve is quite steep, better to start now as there is an explosion in research and advancement in the field. Late entry will make the journey even rockier as the knowledge area expands exponentially. Refer to the above image; the learning curve will be steeper as time goes on.
  4. Community is getting bigger every day; thus, help is around when someone is stuck.
  5. Quantum machines are getting powerful every day. It looks brighter every day for complex computation that any supercomputer or any combination of GPU/TPU would take thousands of years to achieve. But, at the same time, Quantum computer would do it within a few mins.
  6. Many Open-source software development tools are available for Quantum Computation, simulation, and even connect to a quantum computer for proof of concept.
  7. Early research shows QML will outperform CML in many complex machine learning task. For example, the diagram below shows Quantum Neural network can have higher effective dimensions, lower losses faster, and avoid barren plateaus.
Image from YouTube Video by Amira Abbas
Image from YouTube Video by Amira Abbas

What should one learn before Quantum Machine Learning

Image by Author
Image by Author
  1. Linear Algebra over the complex number
  2. Basic probability and number theory
  3. Fourier transformation and its quantum version
  4. Python would be a good choice as many Quantum Computing open-source libraries & frameworks are available in python.
  5. Classical Machine learning would be a prerequisite to understanding many equivalent or advanced QML algorithms.
  6. Quantum Computing with critical concepts, algorithms and implementation of algorithms using python programming would give the necessary foundation to understand QML.

How should one learn Quantum Machine Learning?

Linear Algebra:

Most of the Quantum Computing book covers linear algebra in complex vector space. I mainly found a book from Michael Loceff extremely helpful in this journey as it helps build the knowledge bottoms-up (A Course in Quantum Computing Volume-1 by Michael Loceff). I would strongly recommend keeping this for reference. If any topic sounds complex, you can go back and understand from the fundamental theorem and then understand the crux. There is a supporting YouTube video as well for the book with Hollywood style presentation. Finally, Qiskit has an online textbook which too an excellent introduction to required linear algebra. Linear Algebra for Quantum Computation from Springer contains a concise summary of key areas of liner algebra needed to understand QC. You may also try the below YouTube video.

Probability & Number Theory

I won’t go into details here as lots of material available on these subjects.

Fourier Transformation

Again I would recommend going through Michael Loceff book, which is quite thorough and goes from Fourier series, discrete & Fast Fourier Transform and finally to Quantum Fourier Transformation. YouTube also has tons of material on this topic.

Classical Machine Learning

One must be pretty comfortable with Classical Machine Learning theories, algorithms, different libraries and implementations. Tons of material available in Medium, MOOC, YouTube videos etc.

Quantum Computation

Bible for Quantum Computation is a book by Nielsen and Chaung. Even after two decades, it’s still the best book on the subject despite many sections now a bit dated.

Quantum Computation and Quantum Information: 10th Anniversary Edition

Loceff book also covers basic Quantum Computation concepts and powerful algorithms. I would also recommend few courses like the Umesh Vazirani course from Edx ( you can also find videos on YouTube).

Quantum Mechanics and Quantum Computation

I will also recommend video course content on YouTube from John Preskill.

Please refer to the below pages, which have a curated list of videos, books, MOOC, blog etc., for Quantum Computation.

https://qosf.org/learn_quantum/#massive-open-online-courses

https://github.com/desireevl/awesome-quantum-computing

Python Libraries

Industry giants like IBM, Google back many software tool and development libraries. You will find a comprehensive list and details here. I found Qiskit as most popular among all and easy to use, but like all other software libraries, opinion differs, and you may like Cirq from Google or PyQuil from Rigetti. Qiskit has a lot of content, including a textbook and YouTube channel, many great videos.

Quantum Machine Learning

Finally, we are in the Quantum Machine Learning space. We are in NISQ (Noisy Intermediate-Scale Quantum) era, and it’s important to understand what we can do in the near term with imperfect Quantum machines.

One MOOC is available from Edx with a hands-on introduction to quantum computing and quantum-enhanced machine learning, as well as a code repo.

There are two books on the QML, but none of them I found really valuable as both lack depth in QML areas while spending most of the effort explaining QC.

  • Quantum Machine Learning: What Quantum Computing Means to Data Mining by Peter Wittek
  • Supervised Learning with Quantum Computers by Maria Schuld & Francesco Petruccione

I will also recommend referring QML content and libraries from Penny Lane. This is because PennyLane supports all the major Quantum frameworks.

Image from PennyLane
Image from PennyLane

This Github repository has a comprehensive list of Algorithm for QML.

Qiskit page for machine learning also contains valuable implementation information.

I have created a YouTube playlist for QML, and you will find a wealth of freely available QML content from YouTube in this playlist.

One important suggestion, don’t go into a research paper till you are comfortable with basic concepts and notation, else you will find it difficult to go through the papers.

Final Note

QC and QML both are booming fields with lots of hype, hope and sometimes unreasonable expectation. I have tried to share my experience and the journey that I have taken so far. I would like to hear from you if you have any suggestions or come across any other content or information that can help our thriving community.

Thanks for reading. You can connect me on LinkedIn.

For only $5/month, get unlimited access to the most inspiring and uplifting content… Click on the link below to become a Medium member and support my writing. Thank you! https://baijayanta.medium.com/membership

You may also like to take a look at the below article on Quantum Data Encoding.

All about Data Encoding for Quantum Machine Learning


Related Articles