Dual-VaDE: Efficient and Robust Prototype Learning via Variational Dual-Domain Mixture Priors
Organizations: College of Engineering, Northeastern University, Boston, MA, USA · Computer Science, University of Central Florida, Orlando, FL, USA · Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA
Abstract
In this paper, we propose Dual-VaDE, an unsupervised generative framework for learning interpretable anatomical prototypes from unlabeled medical images. In contrast to previous variational methods that rely on a single spatial latent code, which often entangles global geometry with local textures and fails to capture their cross-domain dependencies, Dual-VaDE introduces a dual-domain Gaussian mixture prior that explicitly couples spatial anatomy with phase-preserving Fourier spectral representations. We derive an exact, closed-form variational lower bound that fuses this joint prior into the decoder and optimizes the entire network end-to-end via an alternating EM formulation, avoiding heuristic two-stage clustering while ensuring superior training stability and robustness. Evaluated on OASIS-3 brain MRI, Dual-VaDE delivers substantially sharper and more plausible prototypes, achieves superior cluster separation with improvements of up to 94% over spatial-only baselines, exhibits markedly reduced variance across random seeds, and incurs negligible inference overhead (0.28 ms per image).
Figures & tables
| Silhouette | Davies-Bouldin | ||||||||
| Model | Spectral input | Phase | Avg. rank | ||||||
| VaDE [ 12 ] | None | 0.022 ‡ | 0.009 ‡ | 0.065 | 0.083 ‡ | 0.252 † | 0.196 | 6.67 | |
| Freq-VaDE | FullFreq [ 8 ] | ✗ | 0.003 † | 0.023 ‡ | 0.039 ‡ | 0.006 | 0.024 ‡ | 0.101 † | 4.33 |
| LowFreq [ 21 ] | ✗ | 0.021 | 0.013 | 0.046 ‡ | 0.022 | 0.011 | 0.348 ‡ | 4.25 | |
| PolarFreq [ 9 ] | ✓ | 0.006 ‡ | 0.036 ‡ | 0.055 ‡ | 0.040 ‡ | 0.034 ‡ | 0.086 ‡ | 8.33 | |
| ReIm [ 24 ] | ✓ | 0.036 † | 0.032 ‡ | 0.049 ‡ | 0.029 ‡ | 0.029 ‡ | 0.310 † | 6.17 | |
Prototype Learning
Prototype learning represents data with representative examples or feature centers to support classification, adaptation, anomaly detection, and interpretable predictions.
See all 88 papers