With a major leap into the future of computing, researchers have shown that even small quantum processors can outperform classical algorithms in machine learning tasks.
This discovery provides a glimpse into a faster, greener age in a relatively new field of research in quantum machine learning.
This new study combines two of the most disruptive technologies of our time: quantum computing and machine learning.
When a photon covers the code
Recent advances in both areas are reshaping the frontier of technology. AI is already embedded in everything from personal assistants to scientific research, but quantum computing promises fundamentally new ways of processing. Their intersection has produced quantum machine learning, a rapidly growing field.
This new field explores whether quantum systems can improve the speed, accuracy, or efficiency of machine learning algorithms. However, demonstrating such benefits to today's limited quantum hardware remains a major challenge, and researchers are just beginning to work on it.
The experiment, conducted by an international team led by the University of Vienna, used Photonic quantum processors to classify data points, an essential task for modern AI systems.
Researchers discovered that quantum systems are better than classic counterparts, resulting in fewer errors. This is a rare glimpse into the quantum advantages of today's hardware.
This breakthrough was achieved using quantum photonic circuits developed at Politecnico di Milano in Italy and machine learning algorithms proposed by UK-based Quantinuum. This experiment is one of the first demonstrations of quantum enhancement in real AI tasks, rather than simulations.
Greener, faster, smarter AI
By separating quantum contributions in the classification process, the team was able to identify specific scenarios in which quantum systems were superior.
Their results not only validate the potential of photonic quantum processors, but also lay the foundation for identifying machine learning tasks that quantum computing can have real impacts, even on today's limited scale hardware.
“For certain tasks, we found that the algorithms have fewer errors than the classic counterparts,” says Philip Walster, project leader at the University of Vienna.
“This means that existing quantum computers can perform well without necessarily surpassing cutting-edge technologies,” adds Zhenghao Yin, the first author of the study.
Beyond accuracy, the experiment also reveals another important advantage in energy efficiency.
Photonic Quantum Systems uses light to process information, which consumes significantly less power than traditional hardware. This has become increasingly important as AI energy demand continues to grow.
“This could prove to be important in the future given that machine learning algorithms are becoming unfeasible due to high energy needs,” said co-author Iris Agresty.
By showing that today's quantum devices can already offer concrete improvements, our findings can guide both quantum computing and classical machine learning into a more symbiotic future.
