Research map / Probabilistic and scientific modeling / Causal discovery
Provable Subspace Identification Under Post-Nonlinear Mixtures
Qi Lyu, Xiao Fu
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 |
|---|---|---|---|
| Increase network width R from 8 to 256 (ablation) | Proposed PNL method · Synthetic, N=10000 | subspace distance: 0.47 -> 0.03 | Improved |
| Increase network width R from 256 to 1024 (ablation) | Proposed PNL method · Synthetic, N=10000 | subspace distance: 0.03 -> 0.07 | Hurt |
| Double sample size N from 10000 to 20000 (ablation) | Proposed PNL method, R=256 · Synthetic | subspace distance: 0.03 -> 0.02 | Improved |
| Double sample size N from 10000 to 20000 (ablation) | Proposed PNL method, R=8 · Synthetic | subspace distance: 0.47 -> 0.44 | No clear effect |
| Use proposed PNL nonlinearity removal before PCA instead of raw data | SVM on EEG · EEG | classification error: 0.46 -> 0.40 | Improved |
| Use proposed PNL nonlinearity removal before PCA instead of raw data | LR on EEG · EEG | classification error: 0.47 -> 0.39 | Improved |
| Replace linear PCA with proposed method | SVM on EEG · EEG | classification error: 0.43 -> 0.40 | Improved |
| Replace autoencoder with proposed method | LR on EEG · EEG | classification error: 0.46 -> 0.39 | Improved |
| Replace nonlinear ICA with proposed method | SVM on EEG · EEG | classification error: 0.44 -> 0.40 | Improved |
| Replace PNL-ICA with proposed method | LR on EEG · EEG | classification error: 0.44 -> 0.39 | Improved |
| Dependent latent components on probability simplex violate sufficient identifiability condition | Proposed PNL method · Synthetic dependent, K=3 M=5 | nonlinearity removal | No clear effect |
Closest papers by meaning
- Diverse Influence Component Analysis: A Geometric Approach to Nonlinear Mixture Identifiability (NeurIPS 2025 · similarity 0.87)
- On the Identifiability of Nonlinear ICA: Sparsity and Beyond (NeurIPS 2022 · similarity 0.85)
- On Finite-Sample Identifiability of Contrastive Learning-Based Nonlinear Independent Component Analysis (ICML 2022 · similarity 0.85)
- Robustness of Nonlinear Representation Learning (ICML 2024 · similarity 0.84)
- Heterogeneous Sufficient Dimension Reduction and Subspace Clustering (ICML 2025 · similarity 0.84)
- Nonlinear Laplacians: Tunable principal component analysis under directional prior information (NeurIPS 2025 · similarity 0.83)
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.
