Medical Imaging Classification with Cold-Atom Reservoir Computing using Auto-Encoders and Surrogate-Driven Training
This work introduces a hybrid quantum-classical pipeline called QGARS for medical image classification, specifically polyp detection. It combines a classical autoencoder with a neutral-atom reservoir computing system and uses a differentiable surrogate model to overcome training difficulties caused by non-differentiable quantum measurements. The method outperforms traditional classical feature reduction techniques.