Research map / Label-efficient and robust learning / Self-supervised and contrastive
Sample and Predict Your Latent: Modality-free Sequential Disentanglement via Contrastive Estimation
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 |
|---|---|---|---|
| Use SPYL's VAE-based sampling and contrastive estimation | Sequential disentanglement · MUG | Acc: 81.16% -> 85.71% | Helped |
| Use SPYL's VAE-based sampling and contrastive estimation | Sequential disentanglement · TIMIT | static EER: 4.03% -> 3.41% | Helped |
| Use SPYL's VAE-based sampling and contrastive estimation | Sequential disentanglement · Physionet ICU mortality | AUROC: 0.752 ± 0.011 -> 0.764 ± 0.040 | Helped |
| Use SPYL's VAE-based sampling and contrastive estimation | Sequential disentanglement · Air Quality average daily rain | MAE: 1.806 ± 0.012 -> 1.823 ± 0.001 | Hurt |
| Sample negative views from the farthest third rather than randomly (ablation) | SPYL · MUG | Acc: 84.18% -> 85.71% | Helped |
| Sample negative views from the farthest third rather than the middle third (ablation) | SPYL · MUG | Acc: 84.43% -> 85.71% | Helped |
| Sample negative views from the farthest third rather than randomly (ablation) | SPYL · TIMIT | EER gap: 28.53% -> 29.81% | Helped |
| Sample negative views from the farthest third rather than the middle third (ablation) | SPYL · TIMIT | EER gap: 29.11% -> 29.81% | Helped |
| Add predictive sampling instead of using a simple reparametrization trick (ablation) | SPYL · Sprites | IS: 8.865 ± \num{9.98e-4} -> 8.942 ± \num{3.3e-5} | Helped |
| Add predictive sampling instead of using a simple reparametrization trick (ablation) | SPYL · MUG | Acc: 83.93% ± 0.96 -> 85.71% ± 0.9 | Helped |
Closest papers by meaning
- Sequential Disentanglement by Extracting Static Information From A Single Sequence Element (ICML 2024 · similarity 0.90)
- Disentanglement Beyond Static vs. Dynamic: A Benchmark and Evaluation Framework for Multi-Factor Sequential Representations (NeurIPS 2025 · similarity 0.90)
- An Information Criterion for Controlled Disentanglement of Multimodal Data (ICLR 2025 · similarity 0.89)
- Can Diffusion Models Disentangle? A Theoretical Perspective (NeurIPS 2025 · similarity 0.89)
- RePercENT: Scaling Disentangled Representation Learning Beyond Two Modalities (NeurIPS 2026 · similarity 0.88)
- Synergy Between Sufficient Changes and Sparse Mixing Procedure for Disentangled Representation Learning (ICLR 2025 · similarity 0.87)
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.