QCNN for Phase Identification

Quantum Convolutional Neural Networks for quantum phase identification — QHack 2024 (3rd Place + Amazon Braket Prize)

At QHack 2024, we developed and trained Quantum Convolutional Neural Networks (QCNNs) for identifying quantum phases of matter.

Highlights:

  • Trained on a GPU-accelerated quantum circuit simulator for rapid prototyping
  • Evaluated robustness on IonQ’s real quantum hardware through AWS Braket
  • Won 3rd place in the “A Matter of Taste” challenge and the Amazon Braket Prize (Top 3)
  • Featured in PennyLane Highlights

QCNNs are quantum analogs of classical convolutional networks, leveraging translationally invariant circuits to classify quantum states efficiently.

GitHub: ruyi101/QCNN-Phases