FHE-feasible diagnostic CNN
1 conv block (1->4ch, 3x3), avg-pool, dense head, 14x14 input
Part of QSMPC-QKD-QHE-AI-Hybrid,
a quantum-safe orchestration demo. This is the plaintext model for the
medical_fl use case and it runs under real CKKS encryption.
Measured performance
| metric | value |
|---|---|
agreement |
0.75 |
disagreement_on_positives |
0.078947 |
metric_delta |
0.134616 |
metric_delta_pp |
13.4616 |
n_eval |
156 |
student_metric |
0.730769 |
student_params |
3226 |
teacher_metric |
0.865385 |
teacher_params |
11171266 |
Published baselines this is measured against
- Target metric: AUC
- Baseline to beat: 0.901 - ResNet-18 @28px, MedMNIST v2, Yang et al., Scientific Data 10:41 (2023) (AUC 0.901 / ACC 0.863)
- Published ceiling: 0.919 - Google AutoML Vision, MedMNIST v2 (AUC 0.919)
- Companion metric shown alongside:
accuracy- reported together because the aggregate figure can look healthy while the class that matters is not.
Training data
- Dataset: MedMNIST v2 BreastMNIST
- Licence: CC-BY-4.0
- Source: https://medmnist.com/ (licence read 2026-08-03)
780 images at 28x28. The tiny tier; also the source of the FHE-feasible T1 student.
Notes and limitations
Genuine encrypted convolution: im2col + CKKS dot_plain, ReLU/max-pool under MPC.
Honest scope
This model is published as part of a research proof of concept, not as a production system.
Numbers above are what this repository measured on the split described, with the code in
scripts/train/. Where a figure is carried from the literature rather than measured here,
it is labelled as such.