Research map / Probabilistic and scientific modeling / Molecules and proteins
Collaborative Recurrent Autoencoder: Recommend while Learning to Fill in the Blanks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| jointly model recurrent content sequences and collaborative filtering | CRAE · CiteULike and Netflix recommendation | recall: 0.7%~1.9% (a relative boost of 2.0%~16.7%) in CiteULike; 3.5%~6.0% (a relative boost of 5.7%~22.5%) in Netflix | Helped |
| use CRAE instead of CDL | CRAE · CiteULike and Netflix | mAP: 35% for CiteULike; 10% for Netflix | Helped |
| use CRAE instead of the strongest listed BLEU baseline | CRAE · CiteULike | BLEU: 46.60 vs 31.47 | Helped |
| use CRAE instead of the strongest listed BLEU baseline | CRAE · Netflix | BLEU: 48.69 vs 17.17 | Helped |
| replace conventional word dropping with wildcard replacement (ablation) | CRAE · CiteULike | Recall@300: 10.53% -> 12.71% | Helped |
| add denoising rather than use no denoising (ablation) | CRAE · CiteULike | Recall@300: 9.14% -> 12.71% | Helped |
| replace beta-pooling with average pooling (a=1, b=1) (ablation) | CRAE · CiteULike | Recall@300: 12.71 -> 11.62 | Hurt |
| set beta-pooling a=400 and b=311 (ablation) | CRAE · CiteULike | Recall@300: 12.71 | Helped |
| set beta-pooling a=1 and b=10 (ablation) | CRAE · CiteULike | Recall@300: 10.48 | Hurt |
| set beta-pooling a=10 and b=1 (ablation) | CRAE · CiteULike | Recall@300: 10.72 | Hurt |
| move beta-pooling weights toward vectors near the bottleneck instead of using only the middle vector (ablation) | CRAE · CiteULike | Recall@300 | Helped |
| reverse sequence order in the decoder (ablation) | CRAE · Recommendation | recommendation performance | No clear effect |
Closest papers by meaning
- DR-VAE: Debiased and Representation-enhanced Variational Autoencoder for Collaborative Recommendation (AAAI 2025 · similarity 0.88)
- Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations (ICML 2024 · similarity 0.86)
- Generalizing Linear Autoencoder Recommenders with Decoupled Expected Quadratic Loss (ICLR 2026 · similarity 0.86)
- On Efficiency-Effectiveness Trade-off of Diffusion-based Recommenders (NeurIPS 2025 · similarity 0.85)
- Think before Recommendation: Autonomous Reasoning-enhanced Recommender (NeurIPS 2025 · similarity 0.84)
- Sparse Enhanced Network: An Adversarial Generation Method for Robust Augmentation in Sequential Recommendation (AAAI 2024 · similarity 0.84)
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.