Research map / Probabilistic and scientific modeling / Molecules and proteins
Learning Sparse Latent Representations with the Deep Copula Information Bottleneck
Aleksander Wieczorek*, Mario Wieser*, Damian Murezzan, Volker Roth
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 |
|---|---|---|---|
| Apply the copula transformation to DIB | DIB · Artificial data | I(y;t): approximately 6 -> approximately 11 | Improved |
| Apply the copula transformation to DIB | DIB · Artificial data | active latent dimensions: 10 out of 18 dimensions -> two dimensions | Improved |
| Apply the copula transformation to DIB | DIB · Artificial data, test data | I(t;y) | Improved |
| Apply the copula transformation to DIB | DIB · Artificial data, reconstruction | reconstruction quality | Improved |
| Apply the copula transformation to DIB | DIB · Artificial data with outliers | robustness to outliers | Improved |
| Apply the copula transformation to DIB | DIB · Artificial data, training convergence | convergence speed: around iteration no. 1000 | Improved |
| Apply the copula transformation to DIB | DIB · Communities and Crime | active latent dimensions: eight dimensions -> four dimensions | Improved |
| Apply the copula transformation to DIB | DIB · Communities and Crime | I(y;t): p-value of 1.6*10^{-16} | Improved |
| Apply the copula transformation to DIB | DIB · Communities and Crime, latent-space structure | latent disentanglement | Improved |
| Replace the copula transformation with mean-zero, unit-variance normalization (ablation) | cDIB · Artificial data | I(y;t) | Hurt |
| Replace the beta transformation with a gamma transformation in the experiment (ablation) | Artificial data · Artificial data | I(y;t) | Improved |
| Vary the hyperparameter lambda (ablation) | cDIB · Communities and Crime | latent disentanglement: 1.79 and 1897.15 | No clear effect |
Closest papers by meaning
- Gaussian Copula Embeddings (NeurIPS 2022 · similarity 0.87)
- Deep Variational Multivariate Information Bottleneck - A Framework for Variational Losses (JMLR 2025 · similarity 0.85)
- The Information Sieve (ICML 2016 · similarity 0.85)
- Neural Mutual Information Estimation with Vector Copulas (NeurIPS 2025 · similarity 0.83)
- Identifying Metric Structures of Deep Latent Variable Models (ICML 2025 · similarity 0.83)
- Structured IB: Improving Information Bottleneck with Structured Feature Learning (AAAI 2025 · similarity 0.83)
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.
