Quantum data encoding is a crucial step in harnessing the power of noisy intermediate-scale quantum computers. To efficiently encode classical data into a quantum state, it is essential to find shallow quantum circuits that are resilient to noise and errors. Here we propose a transformer-decoder model that generates optimal quantum circuits for data encoding. We found that our model requires only a small amount of training data to successfully encode classical information into a quantum state, even for previously unseen inputs. This indicates its understanding of the global structure of quantum circuits in the training data. Notably, our trained model can find shallower quantum circuits than those provided during training, suggesting its potential to uncover efficient circuit structures that have not been explored before. This work paves the way for the development of robust and scalable quantum algorithms.

Close

PrevNext

View figure in article

• Machine learning
• Natural language processing
• Quantum circuits
• Quantum information processing
• Quantum software
• Quantum state transfer