March 24, 2025
Keio University
TOPPAN Holdings Inc.
Keio University (location: Minato-ku, Tokyo; President: Kohei Ito) and TOPPAN Holdings (headquarters: Bunkyo-ku, Tokyo; President & CEO: Hideharu Maro) have demonstrated the effectiveness of quantum machine learning in the quality inspection of agricultural products using small data, showing that it improves detection accuracy compared to conventional classical machine learning methods. The results of this research were published on March 21, 2025 (JST), in "EPJ Quantum Technology," an international scientific journal on quantum technology.
Highlights of this research
Achieved high-precision anomaly detection with small data: A learning model with high discrimination capability was constructed using a small amount of training data (24 normal and 24 abnormal samples for learning).
Exploration of practical quantum circuits: Identified a quantum kernel structure with excellent image inspection performance by verifying different connection methods between qubits and quantum gate configurations.
Superiority of the quantum kernel: SVMs using a quantum kernel achieved higher discrimination performance (F1 score, AUC) compared to a classical kernel (RBF).
Implementation verification on a quantum computer: Gained insights for practical application by verifying the results from a quantum simulator on an actual quantum computer (IBM Quantum).
For the full press release, please see below.