Research map / Video / Video segmentation and tracking
Disentangled Sequential Autoencoder
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 |
|---|---|---|---|
| Represent sequence content with μ_f rather than dynamics with μ_z | Disentangled Sequential Autoencoder, factorised q · TIMIT speaker verification, dim 16 | EER: 17.84% -> 4.78% | Helped |
| Represent sequence content with μ_f rather than dynamics with μ_z | Disentangled Sequential Autoencoder, factorised q · TIMIT speaker verification, dim 64 | EER: 17.49% -> 4.94% | Helped |
| Use full q instead of factorised q (ablation) | Disentangled Sequential Autoencoder · TIMIT speaker verification, μ_f, dim 16 | EER: 4.78% -> 5.64% | Hurt |
| Use full q instead of factorised q (ablation) | Disentangled Sequential Autoencoder · TIMIT speaker verification, μ_z, dim 16 | EER: 17.84% -> 19.20% | Hurt |
| Use full q instead of factorised q (ablation) | Disentangled Sequential Autoencoder · TIMIT speaker verification, μ_f, dim 64 | EER: 4.94% -> 4.82% | Helped |
| Use full q instead of factorised q (ablation) | Disentangled Sequential Autoencoder · TIMIT speaker verification, μ_z, dim 64 | EER: 17.49% -> 18.89% | Hurt |
| Use stochastic time-local latent transitions instead of deterministic LSTM dynamics (ablation) | Disentangled Sequential Autoencoder · Bouncing-ball sequences with consecutive missing frames | incorrectly reconstructed/predicted pixel fraction | Helped |
| Use one global latent state propagated by an LSTM instead of time-local stochastic latents (ablation) | Deterministic LSTM-f and LSTM-c models · Long, complex bouncing-ball sequences | physical realism | Hurt |
Closest papers by meaning
- S3VAE: Self-Supervised Sequential VAE for Representation Disentanglement and Data Generation (CVPR 2020 · similarity 0.90)
- Deep Generative Video Compression (NeurIPS 2019 · similarity 0.89)
- Unsupervised Learning of Disentangled and Interpretable Representations from Sequential Data (NeurIPS 2017 · similarity 0.88)
- High-Quality Joint Image and Video Tokenization with Causal VAE (ICLR 2025 · similarity 0.88)
- Disentangled Recurrent Wasserstein Autoencoder (ICLR 2021 · similarity 0.88)
- Sequential Disentanglement by Extracting Static Information From A Single Sequence Element (ICML 2024 · 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.