latentbrief
Back to news
Research6d ago

Quantum Computers Learn to Tune Themselves While Running

Google AI Research, NVIDIA Dev Blog1 min brief

In brief

  • Scientists at Google have made a major leap in quantum computing by teaching machines to fix themselves on the fly.
  • Their breakthrough, detailed in Nature, involves using reinforcement learning-like training an AI agent-to adjust thousands of control settings without pausing computations.
    • This is crucial because quantum systems are prone to "drift," causing errors during long runs.
  • Until now, fixing these issues meant stopping the entire process to recalibrate, a huge barrier for practical use.
  • The new method lets the computer tweak its own tuning while it operates, similar to an orchestra adjusting instruments mid-performance without missing a beat.
    • This approach uses quantum error correction (QEC) techniques, including neural networks and specialized decoding algorithms, to detect and correct issues in real time.
  • The advance could pave the way for more reliable, long-term quantum computing, potentially solving complex problems that currently stump even the best systems.
  • Researchers are now focused on scaling this method to larger systems and improving its efficiency.

Terms in this brief

Reinforcement Learning
A type of machine learning where an AI agent learns to make decisions by performing actions and receiving feedback in the form of rewards or penalties. The goal is to maximize cumulative reward over time, often used in game playing, robotics, and autonomous systems.

Read full story at Google AI Research, NVIDIA Dev Blog

More briefs