EMTKD at the Edge: An Adaptive Multi-Teacher Knowledge Distillation for Robust Cardiac MRI Classification
doors 2025 · CEUR Workshop Proceedings 3943, pp. 42-57
Editorial summary
A compact cardiac MRI classifier learns from several larger models whose contributions are weighted adaptively. Training combines transfer between datasets, unlabelled scans and differential privacy. Experiments assess classification accuracy across domains and the effects of adaptive weighting, domain adaptation and learning from unlabelled data.
Bibliographic reference
Use the publisher’s record for the citation format required by your journal or organisation.
Oleksandr Chaban, Eduard Manziuk, Olena Markevych, Sergii Petrovskyi, Pavlo Radiuk. "EMTKD at the Edge: An Adaptive Multi-Teacher Knowledge Distillation for Robust Cardiac MRI Classification". doors 2025 · CEUR Workshop Proceedings 3943, pp. 42-57, 2025.
Test a method against your business task.
Describe the decision your system needs to support and the data you have. We assess whether a method fits, define a prototype and agree how to measure its quality, response time and operating cost.