Research map / Probabilistic and scientific modeling / Variational and generative sampling
Variational Autoencoded Regression: High Dimensional Regression of Visual Data on Complex Manifold
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 |
|---|---|---|---|
| Use VAE-trained latent regression and train the decoder to match regressed latent codes to image responses instead of using R-VAE | Visual data regression · Baseball swings, golf swings, and snatch | SSIM (without background): Baseball: 0.489 -> 0.607; Golf: 0.543 -> 0.707; Snatch: 0.205 -> 0.369 | Helped |
| Use VAE-trained latent regression and train the decoder to match regressed latent codes to image responses instead of using R-VAE | Human pose reconstruction · Actors #1–#4 | SSIM: 0.4849 -> 0.7402; 0.4265 -> 0.6743; 0.5094 -> 0.7295; 0.4954 -> 0.7671 | Helped |
| Use VAE-trained latent regression and decoder alignment instead of C-VAE | Human pose reconstruction with other-actor joints · Human pose reconstruction, other-actor joints | SSIM: 0.4059 -> 0.5227 | Helped |
| Use MOGP instead of the proposed method | Visual data regression · Baseball swings, golf swings, and snatch without background | SSIM: 0.607 -> 0.247; 0.707 -> 0.114; 0.369 -> 0.019 | Hurt |
| Use MOGP instead of the proposed method | Visual data regression · Baseball swings, golf swings, and snatch with background | SSIM: 0.610 -> 0.803; 0.752 -> 0.845; 0.377 -> 0.626 | Mixed |
| Use nearest-neighbor regression in the vanilla VAE latent space instead of proposed latent GP regression | Visual data regression · Baseball swings, golf swings, and snatch | SSIM (without background): 0.607 -> 0.210; 0.707 -> 0.213; 0.369 -> 0.198 | Hurt |
| Increase the latent sampling deviation from +0.5σ to +1.5σ (ablation) | Proposed regression method · Baseball swings, golf swings, and snatch | SSIM: Baseball: 0.6453 -> 0.5307; Golf: 0.7203 -> 0.4422; Snatch: 0.4042 -> 0.3629 | Hurt |
| Increase the latent sampling deviation from +0.5σ to +1.5σ (ablation) | R-VAE · Baseball swings, golf swings, and snatch | SSIM: Baseball: 0.4993 -> 0.3825; Golf: 0.5642 -> 0.2417; Snatch: 0.2700 -> 0.0770 | Hurt |
| Add fine-tuning on observed target data pairs after training on the sequence dataset (ablation) | Proposed regression method · Sports image sequences | Regression quality | Helped |
| Replace proposed regression with C-VAE using randomly sampled latent codes | Human pose reconstruction · Human pose images | SSIM: 0.4849 -> 0.7402; 0.4265 -> 0.6743; 0.5094 -> 0.7295; 0.4954 -> 0.7671 | Hurt |
Closest papers by meaning
- Gaussian Process Prior Variational Autoencoders (NeurIPS 2018 · similarity 0.89)
- Variational autoencoders in the presence of low-dimensional data: landscape and implicit bias (ICLR 2022 · similarity 0.88)
- Shape your Space: A Gaussian Mixture Regularization Approach to Deterministic Autoencoders (NeurIPS 2021 · similarity 0.87)
- Sparse Autoencoders, Again? (ICML 2025 · similarity 0.87)
- A Statistical Analysis of Wasserstein Autoencoders for Intrinsically Low-dimensional Data (ICLR 2024 · similarity 0.86)
- Be a Goldfish: Forgetting Bad Conditioning in Sparse Linear Regression via Variational Autoencoders (ICML 2025 · 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.