Research map / Vision-language and multimodal / Cross-modal learning
A coupled autoencoder approach for multi-modal analysis of cell types
Rohan Gala, Nathan Gouwens, Zizhen Yao, Agata Budzillo, Osnat Penn, Bosiljka Tasic, Gabe Murphy, Hongkui Zeng, Uygar Sümbül
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Increase coupling strength λ (ablation) | Coupled autoencoders on Patch-seq electrophysiology representations · Patch-seq | transcriptomic type prediction accuracy | Improved |
| Increase coupling strength λ (ablation) | Coupled autoencoders on Patch-seq transcriptomic representations · Patch-seq | transcriptomic type prediction accuracy | Hurt |
| Increase coupling strength λ (ablation) | Coupled autoencoders on Patch-seq · Patch-seq | adjusted mutual information between independently clustered modalities | Improved |
| Increase coupling strength λ (ablation) | Coupled autoencoders on Patch-seq · Patch-seq | reconstruction error | Hurt |
| Use 3D CCA representations instead of coupled autoencoder representations (ablation) | Cross-modal cell-type prediction · Patch-seq | cell-type prediction accuracy | Hurt |
| Use conventional per-dimension batch normalization instead of full-covariance normalization (ablation) | Coupled autoencoders · MNIST representation experiment | latent representation dimensionality | Hurt |
| Use full-covariance normalization rather than conventional per-dimension batch normalization (ablation) | Coupled autoencoders · MNIST representation experiment | latent representation collapse | Improved |
| Use mini-batch minimum-singular-value normalization rather than full-covariance normalization (ablation) | Coupled autoencoders · FACS | reconstruction performance | Improved |
| Align representations with coupled autoencoders when only a sparse subset of samples is paired | Cell-type representations from FACS subsets · FACS partially paired subsets | separation of types unique to one subset: p<0.01 | Improved |
| Align representations with coupled autoencoders when shared types have no designated paired samples | Cell-type representations from FACS subsets · FACS partially paired subsets | nearest-neighbor distance consistency for shared type: p=0.89 | No clear effect |
Closest papers by meaning
- Reproducibility of predictive networks for mouse visual cortex (NeurIPS 2024 · similarity 0.86)
- Learning to cluster neuronal function (NeurIPS 2025 · similarity 0.86)
- Multi-Modal Latent Variables for Cross-Individual Primary Visual Cortex Modeling and Analysis (AAAI 2025 · similarity 0.86)
- Multi-modal Gaussian Process Variational Autoencoders for Neural and Behavioral Data (ICLR 2024 · similarity 0.86)
- Identifying interactions across brain areas while accounting for individual-neuron dynamics with a Transformer-based variational autoencoder (NeurIPS 2025 · similarity 0.86)
- Neuroformer: Multimodal and Multitask Generative Pretraining for Brain Data (ICLR 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
