Dr. Adi Makmal brings together quantum physics and learning processes

A quantum computer can solve certain problems much faster – light-years faster – than a standard computer. Dr Adi Makmal attempts to use this technology to solve machine learning problems, and she is looking for partners.

In October 2019, computer search colossus Google reported that its quantum processor, Sycamore, achieved quantum supremacy with 3 minutes and 20 seconds computation that would have taken the strongest classical computer some 10,000 years to complete. “While bits, the basic units of regular computers can be either 1 or 0, qubits (quantum bits) can represent both at the same time. This is one of several quantum properties that allow for parallelization of information processing and, as a result, for faster computing,” explains Dr Adi Makmal. “Hardware that is based on components with quantum properties can perform faster computations for specific problems, such as finding the prime factors of a number. This is significant, because the internet’s encryption systems, for example, are based on an RSA algorithm that assumes that it is impossible to quickly break down a number into prime factors. If quantum computers could expedite this process, the entire internet encryption layout will collapse. Nevertheless, it is important to understand that not all problems can be solved faster using a quantum computer. The challenge is to find the ones for which quantum computing will have a significant advantage over its classical counterpart.”

Makmal (41, married+2) has been trying to identify these problems since her postdoctoral fellowship at the Institute for Quantum Optics and Quantum Information (IQOQI) and the Institute for Theoretical Physics at the University of Innsbruck, Austria. “I study the interface between quantum computing and machine learning, in an attempt to develop learning algorithms that use quantum tools,” she explains. “Machine learning is the ability to take copious amounts of unprocessed data and let an algorithm process it, identify patterns in it and reach conclusions. It allows for the mass processing of data and is relevant for a variety of fields, including economics, physics, chemistry and biology. Machine learning can help with medical data analysis (say, a computer that weighs data concerning a certain symptom, sampled from millions of people), smart cars or recommendation systems – the ones responsible for the recommendations we get on Facebook, e-commerce sites and even search engines.”

After completing her postdoctoral fellowship in 2017, Makmal returned to Israel and joined the Microsoft Israel recommendation team as a machine learning researcher. In October 2019, she joined the Faculty of Engineering at Bar-Ilan as a researcher and is currently establishing her research group, focused on quantum machine learning. “The research is focused on everything pertaining to machine learning, on the one hand, quantum computing, on the other, and the integration of the two – for example, quantum deep learning, which proposes a computational model of neural networks for quantum hardware, or quantum reinforcement learning, which uses quantum algorithms to realize learning based on reinforcements from one’s surroundings. “Of course, it should be noted that at this point, these models are only theoretical. Only a few companies have quantum computers at their disposal, and each one only has several dozens of qubits,” she clarifies. “Still, it is a growing field that can be taken in fascinating directions. Research combining quantum computing and learning has only started over the past decade and is based on the understanding that there are similarities between the mathematical structure of learning problems and quantum computing. Particularly, the mathematical representations of both fields lie within abstract, high-dimensional vector spaces, and both use the same mathematical tools – mainly linear algebra and probability theory. On a personal level, I have always been interested in human learning, both physically and psychologically; at the same time, I’m also interested in quantum physics, the laws that control matter around us. There’s something in this combination, that allows me to explore both - seemingly unrelated – fields, that makes me very happy,” she smiles.

What practical direction can this study take? “My hopes are that the experimentalists will be able to build stronger quantum computers with more qubits and that us theoreticians will develop quantum learning algorithms that will allow fast processing of incredible amounts of data, far more than is possible today. This is the practical motivation for the research. On a day-to-day scale, each step that gets us closer to better understanding machine learning on the one hand, and quantum computing on the other, is a fascinating challenge in itself. I’m looking for curious and hard-working M.Sc. and PhD students from the fields of computer science, engineering, physics and mathematics, who are interested in quantum computing theory and want to take part in this journey.”