Metraipolis › archive › dc_5f45d5857232
CANAL: Channel-Aware Noise Allocation for Differentially Private Feature Distillation in Medical Image Segmentation — arXiv:2609.13271
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.13271
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T23:54:11.835Z
- SHA-256 of the archived text
3f9eed94b8f85eb66ed2cc5ee130af535df521b379da66ed71ab7c6676dd23ee- Stable URL of this page
https://metraipolis.com/documento/dc_5f45d5857232
How to cite
CANAL: Channel-Aware Noise Allocation for Differentially Private Feature Distillation in Medical Image Segmentation — arXiv:2609.13271 — arXiv (en), https://arxiv.org/abs/2609.13271 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5f45d5857232 (Metraipolis archive dc_5f45d5857232, sha256 3f9eed94b8f85eb66ed2cc5ee130af535df521b379da66ed71ab7c6676dd23ee)
BibTeX
@misc{dc5f45d5857232,
title = {CANAL: Channel-Aware Noise Allocation for Differentially Private Feature Distillation in Medical Image Segmentation — arXiv:2609.13271},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.13271},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5f45d5857232; sha256 3f9eed94b8f85eb66ed2cc5ee130af535df521b379da66ed71ab7c6676dd23ee},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.13271. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Medical image segmentation needs diverse training data, but hospitals hold complementary scans they cannot share for privacy and regulatory reasons. Knowledge distillation can bridge this gap by exporting learned feature representations instead of images, but those representations still encode patient-specific anatomy and remain vulnerable to membership-inference and feature-inversion attacks. Adding calibrated Gaussian noise restores a differential-privacy guarantee, yet three issues have been overlooked. First, prior DP feature-distillation pipelines re-sample noise at every student iteration, so each patient image is released many times and the privacy cost composes over those releases, growing by orders of magnitude. We present a sample-once-per-image release, realized by a single precomputation pass, under which each patient contributes one release. Second, uniform noise is wasteful because channels differ in task importance. Using task-gradient energy as the importance measure, we derive CANAL, a closed-form water-filling allocation that gives important channels proportionally less noise, and prove it strictly minimizes importance-weighted distortion at a fixed budget. Third, the clipping caps and importance scores that drive the allocation are themselves data-dependent, so releasing them in the clear silently breaks the guarantee. We give a DP-honest budget split that charges each to the privacy budget, so the reported epsilon is the true epsilon. Across three medical segmentation benchmarks spanning dermoscopy, colonoscopy, and ultrasound, CANAL retains more task-relevant signal than uniform noise at the same privacy budget. Tema degli esploratori: fiducia digitale e crittografia (retro-catalogo arXiv: annunci del 2026-09-15 in cs.CR). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Cryptography and Security (cs.CR); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.13271 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-15T23:20:42.729Z dalla pagina https://arxiv.org/catchup/cs.CR/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).