# Unsupervised learning of acquisition variability in structural connectomes via hybrid latent space modeling

arXiv: https://arxiv.org/abs/2605.13933 · HTML: https://mlsanity.com/paper/2605.13933/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 38 references that are in the map.
- First posted on arXiv 13 May 2026.

## Builds on (most cited first, 8 of 8)

- [Understanding disentangling in β-VAE](https://mlsanity.com/paper/1804.03599/) (2018)
- [Learning Disentangled Joint Continuous and Discrete Representations](https://mlsanity.com/paper/1804.00104/) (2018)
- [3D Whole Brain Segmentation using Spatially Localized Atlas Network Tiles](https://mlsanity.com/paper/1903.12152/) (2019)
- [HACA3: A Unified Approach for Multi-site MR Image Harmonization](https://mlsanity.com/paper/2212.06065/) (2022)
- [Graph Autoencoders for Embedding Learning in Brain Networks and Major Depressive Disorder Identification](https://mlsanity.com/paper/2107.12838/) (2021)
- [Disentangling A Single MR Modality](https://mlsanity.com/paper/2205.04982/) (2022)
- [Structural Connectome Harmonization Using Deep Learning: The Strength of Graph Neural Networks](https://mlsanity.com/paper/2507.13992/) (2025)
- [Characterizing Continuous and Discrete Hybrid Latent Spaces for Structural Connectomes](https://mlsanity.com/paper/2512.02032/) (2025)

## Abstract (arXiv)

Acquisition differences across sites, scanners, and protocols in dMRI introduce variability that complicates structural connectome analysis. This motivates deep learning models that can represent high-dimensional connectomes in a low-dimensional space while explicitly separating acquisition-related effects from biological variation. Conventional dimensionality reduction methods model all variance as continuous, so acquisition effects often get absorbed into a continuous latent space. Recent hybrid latent-space models combine discrete and continuous components to address this, but typically require manual capacity tuning to ensure the discrete component captures the intended variability. We introduce an unsupervised framework that removes this manual tuning by architecturally annealing encoder outputs before decoding, allowing the model to adaptively balance discrete and continuous latent variables during training. To evaluate it, we curated a dataset of N=7,416 structural connectomes derived from dMRI, spanning ages 2 to 102 and 13 studies with 25 unique acquisition-parameter combinations. Of these, 5,900 are cognitively unimpaired, 877 have mild cognitive impairment (MCI), and 639 have Alzheimer's disease (AD). We compare against a standard VAE, PCA with k-means clustering, and hybrid models that anneal only through the loss function. Our architectural annealing produces stronger site learning (ARI=0.53, p<0.05) than these baselines. Results show that a hybrid continuous-discrete latent space, with architectural rather than loss-based annealing, provides a useful unsupervised mechanism for capturing acquisition variability in dMRI: by jointly modeling smooth and categorical structure, the Joint-VAE recovers clusters aligned with scanner and protocol differences.