New algorithm boosts multitasking in quantum machine learning

For a quantum computer to simulate dynamic processes or process data, among other basic tasks, it must convert complex input data into “quantum data” that it can understand. This process is called quantum compilation.

Essentially, quantum compilation “programs” a quantum computer by converting specific targets into executable sequences. Just as a GPS app converts your desired destination into a series of actionable steps you can follow, quantum compilation converts high-level goals into a precise sequence of quantum operations that a quantum computer can perform.

Traditionally, quantum compilation algorithms optimize a single target at a time. While this approach is effective, it has its limitations. Many complex applications require quantum computers to perform multiple tasks. For example, when simulating quantum dynamic processes or preparing quantum states for experiments, researchers may need to manage multiple operations simultaneously to obtain accurate results. In these situations, it becomes inefficient to deal with one target at a time.

To address these challenges, a team led by Dr. Le Bin Ho of Northeastern University developed a multi-objective quantum compilation algorithm. They published new research in the journal “Machine Learning: Science and Technology” on December 5, 2024.

“By enabling quantum computers to optimize multiple objectives simultaneously, the algorithm increases flexibility and maximizes performance,” Le said. This leads to improvements in the simulation of complex systems or tasks involving multiple variables in quantum machine learning, making it ideal for applications across different scientific disciplines.

In addition to performance improvements, this multi-objective algorithm opens the door to new applications previously limited by single-objective methods. In materials science, for example, researchers can use the algorithm to simultaneously explore multiple properties of materials at the quantum level. In physics, this algorithm can help study systems that are constantly evolving or require a full understanding of various interactions.

This development represents a major advance in quantum computing. Le added: “The multi-objective quantum compilation algorithm brings us one step closer to the day when quantum computers can efficiently handle complex, multi-faceted tasks and provide solutions to problems that cannot be solved by classical computers.”

Going forward, Le aims to study how the algorithm adapts to various types of noise and find ways to improve its performance.