Research map / Learning theory and algorithms / Matrix and tensor methods
Spectral Approximate Inference
Sejun Park, Eunho Yang, Se-Young Yun, Jinwoo Shin
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 |
|---|---|---|---|
| Approximate the partition function using spectral decomposition and spectral mean-field factorization | General high-rank graphical models · Synthetic high-rank GMs, especially high-coupling settings | log partition error | Improved |
| Approximate using spectral inference instead of BP, mean-field, or mini-bucket elimination | Synthetic high-rank graphical models · High-coupling region | log partition error | Improved |
| Use spectral inference instead of BP, MF, MBE, or WMBE | UAI 2014 grid GMs · UAI 2014 grid GMs | log partition error | Improved |
| Use the low-rank spectral algorithm | Rank-1 graphical models · Rank-1 GMs, including antiferromagnetic cases | log partition error | Improved |
| Optimize the diagonal with the SDP rather than use D=0 (ablation) | Spectral mean-field inference · Complete graph GMs, 20 vertices | log partition error | Improved |
| Optimize the diagonal with the SDP rather than use D=-max_j lambda_j times I (ablation) | Spectral mean-field inference · Complete graph GMs, 20 vertices | log partition error | Improved |
| Optimize the diagonal with the SDP rather than use a diagonal-dominance diagonal (ablation) | Spectral mean-field inference · Complete graph GMs, 20 vertices | log partition error | Improved |
| Run spectral inference instead of BP | Inference on complete graphs of varying size · Complete graph GMs, 100-500 vertices | running time | No clear effect |
Closest papers by meaning
- Graphical model inference: Sequential Monte Carlo meets deterministic approximations (NeurIPS 2018 · similarity 0.87)
- Gauging Variational Inference (NeurIPS 2017 · similarity 0.86)
- Bucket Renormalization for Approximate Inference (ICML 2018 · similarity 0.85)
- Stochastic Discrete Clenshaw-Curtis Quadrature (ICML 2016 · similarity 0.85)
- Inference in Graphical Models via Semidefinite Programming Hierarchies (NeurIPS 2017 · similarity 0.85)
- EigenVI: score-based variational inference with orthogonal function expansions (NeurIPS 2024 · similarity 0.84)
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.
