Research map / Probabilistic and scientific modeling / Variational and generative sampling
Non-linear, Sparse Dimensionality Reduction via Path Lasso Penalized Autoencoders
Oskar Allerbo, Rebecka Jörnsten
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 path lasso instead of parameter-wise lasso | sparse autoencoder · Synthetic 16 clusters (noise) | explained variance R2: 0.58 -> 0.89 | Improved |
| use path lasso instead of sparse PCA | sparse dimensionality reduction · Synthetic 16 clusters (noise) | explained variance R2: 0.50 -> 0.89 | Improved |
| use path lasso instead of dense autoencoder | autoencoder · Synthetic 16 clusters (noise) | explained variance R2: 0.86 -> 0.89 | Improved |
| use path lasso instead of standard lasso | sparse autoencoder · Synthetic 16 clusters (no noise) | explained variance R2: 0.98 -> 0.98 | No clear effect |
| use non-linear autoencoder instead of linear PCA | dimensionality reduction · Synthetic 16 clusters (no noise) | explained variance R2: 0.48 -> 0.98 | Improved |
| use path lasso instead of standard lasso | sparse autoencoder · 20 Newsgroups high sparsity (2 dims) | explained variance R2: 0.015 -> 0.13 | Improved |
| use path lasso instead of sparse PCA | sparse PCA · 20 Newsgroups high sparsity (4 dims) | explained variance R2: 0.11 -> 0.20 | Improved |
| use path lasso instead of dense autoencoder | autoencoder · 20 Newsgroups low sparsity (4 dims) | explained variance R2: 0.33 -> 0.34 | Improved |
| increase latent dimensionality from 2 to 25 | non-linear vs linear models · 20 Newsgroups | explained variance advantage: 0.13 -> 0.55 | Hurt |
| use path lasso instead of dense autoencoder | autoencoder · AT&T faces low sparsity | observation reconstruction match: 0.66 -> 0.70 | Improved |
| use path lasso instead of standard lasso | sparse autoencoder · AT&T faces high sparsity | 10-NN reconstruction match: 0.26 -> 0.75 | Improved |
| use path lasso instead of sparse PCA | sparse PCA · AT&T faces high sparsity | explained variance R2: -1.2 -> 0.57 | Improved |
Closest papers by meaning
- The tree autoencoder model, with application to hierarchical data visualization (NeurIPS 2024 · similarity 0.87)
- Be a Goldfish: Forgetting Bad Conditioning in Sparse Linear Regression via Variational Autoencoders (ICML 2025 · similarity 0.86)
- SASA: Subspace-Aware Sparse Autoencoders for Effective Mechanistic Interpretability (NeurIPS 2026 · similarity 0.86)
- Compression of Structured Data with Autoencoders: Provable Benefit of Nonlinearities and Depth (ICML 2024 · similarity 0.85)
- Sparse Autoencoders, Again? (ICML 2025 · similarity 0.85)
- SSC-VAE: Structured Sparse Coding Based Variational Autoencoder for Detail Preserved Image Reconstruction (AAAI 2025 · 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.
