Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Structured Inference Networks for Nonlinear State Space Models
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 |
|---|---|---|---|
| Add edges from x_{t-1} to z_t and x_t | DMM · Polyphonic music: JSB, Nottingham, Piano, Musedata | test negative log-likelihood: 6.388 -> 6.288; 2.770 -> 2.679; 7.835 -> 7.591; 6.831 -> 6.356 | Helped |
| Use future observations and the previous latent state in structured inference (DKS) rather than past-only structured inference (ablation) | DMM inference network · JSB | test negative log-likelihood: 7.020 -> 6.605 | Helped |
| Use future observations and the previous latent state in structured inference (DKS) rather than past-only structured inference (ablation) | DMM inference network · Nottingham | test negative log-likelihood: 3.446 -> 3.136 | Helped |
| Use future observations and the previous latent state in structured inference (DKS) rather than past-only structured inference (ablation) | DMM inference network · Piano | test negative log-likelihood: 9.375 -> 8.471 | Helped |
| Use future observations and the previous latent state in structured inference (DKS) rather than past-only structured inference (ablation) | DMM inference network · Musedata | test negative log-likelihood: 8.301 -> 7.280 | Helped |
| Use DKS instead of structured inference conditioned on both past and future observations (ablation) | DMM inference network · JSB | test negative log-likelihood: 6.632 -> 6.605 | Helped |
| Use DKS instead of structured inference conditioned on both past and future observations (ablation) | DMM inference network · Nottingham | test negative log-likelihood: 3.251 -> 3.136 | Helped |
| Use DKS instead of structured inference conditioned on both past and future observations (ablation) | DMM inference network · Piano | test negative log-likelihood: 8.406 -> 8.471 | Hurt |
| Use DKS instead of structured inference conditioned on both past and future observations (ablation) | DMM inference network · Musedata | test negative log-likelihood: 7.127 -> 7.280 | Hurt |
| Add dependence on the previous latent state to mean-field inference using future observations (ablation) | DMM inference network · JSB | test negative log-likelihood: 6.701 -> 6.605 | Helped |
| Add dependence on the previous latent state to mean-field inference using future observations (ablation) | DMM inference network · Nottingham | test negative log-likelihood: 3.273 -> 3.136 | Helped |
| Add dependence on the previous latent state to mean-field inference using future observations (ablation) | DMM inference network · Piano | test negative log-likelihood: 9.188 -> 8.471 | Helped |
Closest papers by meaning
- eXponential FAmily Dynamical Systems (XFADS): Large-scale nonlinear Gaussian state-space modeling (NeurIPS 2024 · similarity 0.87)
- Gated Inference Network: Inference and Learning State-Space Models (NeurIPS 2024 · similarity 0.87)
- Structural Inference of Dynamical Systems with Conjoined State Space Models (NeurIPS 2024 · similarity 0.87)
- Understanding and Mitigating Bottlenecks of State Space Models through the Lens of Recency and Over-smoothing (ICLR 2025 · similarity 0.86)
- State-Free Inference of State-Space Models: The *Transfer Function* Approach (ICML 2024 · similarity 0.86)
- Theoretical Foundations of Deep Selective State-Space Models (NeurIPS 2024 · 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.