Research map / Probabilistic and scientific modeling / Molecules and proteins
Gaussian Copula Embeddings
Chien Lu, Jaakko Peltonen
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 |
|---|---|---|---|
| Model anime rating and watched episodes jointly with GCE instead of p-emb | GCE · Anime ratings held-out | MAE: 3.0857 -> 1.2170 (K=50) | Improved |
| Use GCE instead of Poisson matrix factorization | GCE · Anime ratings held-out | MAE: 7.4866 -> 1.2170 (K=50) | Improved |
| Increase embedding dimension from K=50 to K=100 (ablation) | GCE · Anime ratings held-out | MAE: 1.2170 -> 1.2207 | No clear effect |
| Model kills and deaths jointly with GCE instead of separate normal embeddings | GCE · HLTV match records | MAE (Kills): 30.0125 -> 14.0867 (K=50) | Improved |
| Model kills and deaths jointly with GCE instead of separate Poisson embeddings | GCE · HLTV match records | MAE (Deaths): 32.5620 -> 13.5245 (K=50) | Improved |
| Increase embedding dimension from K=50 to K=100 (ablation) | GCE · HLTV match records | MAE (Kills): 14.0867 -> 14.0766 | No clear effect |
| Add GCE learned source-IP embeddings as features to random forest | Random forest · CIC Darknet traffic | accuracy: 0.8917 -> 0.9355 | Improved |
| Add GCE learned features to random forest | Random forest · CIC Darknet traffic | F1: 0.8117 -> 0.8851 | Improved |
| Random forest with GCE features instead of DeepImage CNN | GCE (K = 30) + Random forest · CIC Darknet traffic | accuracy: 0.86 -> 0.9355 | Improved |
| Vary GCE embedding dimension K=20, 30, 40 (ablation) | GCE + Random forest · CIC Darknet traffic | accuracy: 0.9347 -> 0.9355 -> 0.9346 | No clear effect |
| Use GCE with node meta-data instead of Deepwalk | GCE · Spanish Twitch gamers link prediction | AUC: 0.7151 -> 0.7853 (K=50) | Improved |
| Use GCE instead of Node2vec at higher dimension | GCE · Spanish Twitch gamers link prediction | AUC: 0.6612 -> 0.7832 (K=100) | Improved |
Closest papers by meaning
- Learning Sparse Latent Representations with the Deep Copula Information Bottleneck (ICLR 2018 · similarity 0.87)
- Learning to Embed Distributions via Maximum Kernel Entropy (NeurIPS 2024 · similarity 0.84)
- Neural Mutual Information Estimation with Vector Copulas (NeurIPS 2025 · similarity 0.83)
- Cross-Modal Alignment via Variational Copula Modelling (ICML 2025 · similarity 0.82)
- Generalization Guarantees for Representation Learning via Data-Dependent Gaussian Mixture Priors (ICLR 2025 · similarity 0.82)
- Optimization Can Learn Johnson Lindenstrauss Embeddings (NeurIPS 2024 · similarity 0.81)
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.
