Research map / Probabilistic and scientific modeling / Variational and generative sampling
Learning latent variable structured prediction models with Gaussian perturbations
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 |
|---|---|---|---|
| Replace exact structured-output search during training with randomized structured outputs, retaining exact test inference | Latent-variable structured prediction · Directed spanning trees, directed acyclic graphs, and cardinality-constrained sets | test distortion: 21% ± 2.6% -> 15% ± 1.6%; 20% ± 1.7% -> 19% ± 1.5%; 20% ± 3.1% -> 16% ± 2.1% | Helped |
| Replace exact structured-output search during training with randomized structured outputs, retaining exact test inference | Latent-variable structured prediction · Directed spanning trees, directed acyclic graphs, and cardinality-constrained sets | test distortion: 15% ± 2.8% vs 15% ± 1.6%; 18% ± 1.8% vs 19% ± 1.5%; 15% ± 2.8% vs 16% ± 2.1% | No clear effect |
| Use randomized rather than exact structured-output search during training | Latent-variable structured prediction · Directed spanning trees, directed acyclic graphs, and cardinality-constrained sets | training runtime: 1024 ± 5 -> 33 ± 0; 1024 ± 4 -> 51 ± 1; 1020 ± 4 -> 51 ± 0 | Helped |
| Use randomized structured outputs during training but random approximate inference at test time | Latent-variable structured prediction · Directed spanning trees | test distortion: 21% ± 2.6% -> 30% ± 1.5% | Hurt |
| Use randomized structured outputs during training but random approximate inference at test time | Latent-variable structured prediction · Directed acyclic graphs | test distortion: 20% ± 1.7% -> 25% ± 1.2% | Hurt |
| Use randomized structured outputs during training but random approximate inference at test time | Latent-variable structured prediction · Cardinality-constrained sets | test distortion: 20% ± 3.1% -> 31% ± 1.7% | Hurt |
| Apply randomized learning with latent variables to image keypoint matching | Image matching · Buffy Stickmen (test) | matching error: 8.69 -> 0.3878 | Helped |
Closest papers by meaning
- Learning Randomly Perturbed Structured Predictors for Direct Loss Minimization (ICML 2021 · similarity 0.88)
- Leveraging Recursive Gumbel-Max Trick for Approximate Inference in Combinatorial Spaces (NeurIPS 2021 · similarity 0.88)
- Efficient Marginalization of Discrete and Structured Latent Variables via Sparsity (NeurIPS 2020 · similarity 0.87)
- A Consistent Regularization Approach for Structured Prediction (NeurIPS 2016 · similarity 0.87)
- On Certified Generalization in Structured Prediction (NeurIPS 2023 · similarity 0.87)
- Amortized Structured Stochastic Variational Inference for Gaussian Process Latent Variable Models (NeurIPS 2026 · 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.