Research map / Learning theory and algorithms / Kernels and regression theory
High-dimensional Asymptotics of Denoising Autoencoders
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 |
|---|---|---|---|
| include a skip connection (full DAE) instead of bottleneck network alone (ablation) | two-layer tied-weight DAE · Gaussian mixture denoising | denoising test MSE | Helped |
| compare the full DAE with PCA denoising | DAE · Gaussian mixture denoising | denoising test MSE | Helped |
| add the bottleneck network to the learnt rescaling component (ablation) | rescaling-only denoiser · Gaussian mixture, low noise levels | denoising test MSE | Helped |
| add the bottleneck network to the rescaling component at high noise (ablation) | rescaling-only denoiser · Gaussian mixture, high noise levels | denoising test MSE | No clear effect |
| increase sample complexity alpha | DAE · isotropic binary mixture | gap of MSE to oracle denoiser | Helped |
| use a non-linear activation instead of a linear model (ablation) | DAE · Gaussian mixture denoising | denoising test MSE | Helped |
Closest papers by meaning
- Denoising Criterion for Variational Auto-Encoding Framework (AAAI 2017 · similarity 0.88)
- Noisier2Noise: Learning to Denoise From Unpaired Noisy Data (CVPR 2020 · similarity 0.87)
- Noise2Self: Blind Denoising by Self-Supervision (ICML 2019 · similarity 0.87)
- Sublinear Variational Optimization of Gaussian Mixture Models with Millions to Billions of Parameters (NeurIPS 2026 · similarity 0.87)
- Compression of Structured Data with Autoencoders: Provable Benefit of Nonlinearities and Depth (ICML 2024 · similarity 0.86)
- Analysis of Learning a Flow-based Generative Model from Limited Sample Complexity (ICLR 2024 · similarity 0.85)
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.