Skip to content

Disentangled Representation Learning

Definition
AI-generated

Disentangled representation learning trains models so that latent variables (or subspaces) align with distinct generative factors of variation—such as identity, pose, lighting, batch, or modality—rather than entangling them in arbitrary mixed axes.

Topics
Machine learning concepts
Related
Batch Effect Deep Generative Model Foundation Model

Why it matters in GWAS

In genomics and imaging pipelines that feed GWAS follow-up (eQTL atlases, spatial or multi-omic integration, phenotype prediction), disentanglement-oriented models can reduce confounding between biological signal and technical or demographic axes—when assumptions hold and evaluation is careful. Poorly validated “disentanglement” can still absorb genetic structure spuriously.

Example usage

"Quality control and downstream modeling both referenced Disentangled Representation Learning in the analysis workflow."

Batch Effect Deep Generative Model Foundation Model In Silico Perturbation Prediction Multi-Omic Single-Cell Integration
Related-term network

Click any node to open that term page.

Node order. The current term is first. Next come every page linked from Related terms, in the same top-to-bottom order as that list. Then new terms appear in two-hop layers: for each of those direct neighbors (in that same order), their related terms are appended if not already listed. A third pass does the same for two-hop nodes. The walk stops when a fixed maximum number of nodes is reached.

References

  • Bengio Y, Courville A, Vincent P. (2013). Representation learning: a review and new perspectives. IEEE TPAMI.
  • Higgins I, et al. (2017). beta-VAE: learning basic visual concepts with a constrained variational framework. ICLR.
  • Locatello F, et al. (2019). Challenging common assumptions in the unsupervised learning of disentangled representations. ICML.
← Discrete uniform distribution Disruptive selection →

Last updated (UTC · Git history)