Does a Quantum Kernel Classifier Degrade Less Under Chest X-ray Preprocessing Shift? A Leakage-Controlled Classical–Quantum Benchmark

Résumé fourni par la source

This repository contains the reproducibility materials supporting the manuscript “Do Quantum Kernels Degrade Less Under Chest X-ray Preprocessing Shift? A Leakage-Controlled Cross-Pipeline Benchmark.” The archive contains the complete 468-run experiment results, configuration-level and image-level prediction outputs, statistical-analysis outputs, experimental notebook, derived ResNet18 features, quantum-kernel outputs, fine-tuned ResNet18 reference-model outputs and checkpoint, documentation, and reproducibility metadata. The study compares quantum and classical classifiers under a controlled chest X-ray preprocessing/representation shift. The source and target-pipeline data originate from the same underlying Guangzhou pediatric cohort and therefore do not represent independent cross-hospital validation. Legacy filenames containing “external” refer to the target processing pipeline. Original medical images are not redistributed. PneumoniaMNIST and the original Kermany et al. chest X-ray data should be obtained from their respective public sources.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.