Research map / Label-efficient and robust learning / Self-supervised and contrastive
S3VAE: Self-Supervised Sequential VAE for Representation Disentanglement and Data Generation
Check 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 |
|---|---|---|---|
| Add the combined self-supervision and regularization objectives to the S3VAE baseline | S3VAE · MUG | Acc: 53.83% -> 70.51% | Helped |
| Add static consistency constraint (ablation) | S3VAE · MUG | Acc: 61.45% -> 70.51% | Helped |
| Add dynamic factor prediction (ablation) | S3VAE · MUG | Acc: 58.32% -> 70.51% | Helped |
| Add mutual information regularization (ablation) | S3VAE · MUG | Acc: 66.07% -> 70.51% | Helped |
| Use self-supervision rather than ground-truth labels for disentanglement | S3VAE · SMMNIST | Acc: 92.18% -> 95.09% | Helped |
| Use self-supervision rather than ground-truth labels for disentanglement | S3VAE · MUG | Acc: 72.32% -> 70.51% | Hurt |
| Use the dynamic representation z_1:T for speaker verification instead of the static representation z_f | S3VAE · TIMIT | EER: 4.80% -> 40.12% | Helped |
Closest papers by meaning
- Contrastively Disentangled Sequential Variational Autoencoder (NeurIPS 2021 · similarity 0.91)
- Disentangled Sequential Autoencoder (ICML 2018 · similarity 0.90)
- Guided Variational Autoencoder for Disentanglement Learning (CVPR 2020 · similarity 0.89)
- Sample and Predict Your Latent: Modality-free Sequential Disentanglement via Contrastive Estimation (ICML 2023 · similarity 0.89)
- Sequential Disentanglement by Extracting Static Information From A Single Sequence Element (ICML 2024 · similarity 0.88)
- $\alpha$TC-VAE: On the relationship between Disentanglement and Diversity (ICLR 2024 · similarity 0.86)
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.