Research map / Probabilistic and scientific modeling / Dynamical systems
Reduced-Rank Linear Dynamical Systems
Qi She, Yuan Gao, Kai Xu, Rosa Chan
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 |
|---|---|---|---|
| add reduced-rank prior (multivariable Laplacian or nuclear norm) on dynamics matrix (ablation) | Poisson LDS (PLDS) · Simulated counts, true rank 10, 20 latent states | recovered rank of dynamics matrix: rank 10 from 20D state space | Improved |
| add reduced-rank prior plus dispersion-adaptive distribution (RRLDS-ML/-NN) versus PLDS (ablation) | RRLDS · Simulated counts, ranks 2-10 | predictive log-likelihood: p < 0.001 | Improved |
| use dispersion-adaptive distribution instead of Poisson observation (ablation) | PLDS · Simulated counts, ranks 2-10 | predictive log-likelihood | Improved |
| add reduced-rank structure on top of DALDS (ablation) | DALDS · Simulated counts, ranks 2-10 | predictive log-likelihood | Improved |
| add reduced-rank structure when training data is short (ablation) | DALDS · Simulated counts, rank 5, short training length | predictive log-likelihood | Improved |
| increase training data length up to 2000 (ablation) | RRLDS vs DALDS · Simulated counts, rank 5 | predictive log-likelihood | No clear effect |
| RRLDS versus PLDS, SubspaceID, StableLDS and LDS across numbers of latent states | RRLDS-ML/-NN · Hippocampus spiking Task #1 | predictive log-likelihood: p < 0.001 | Improved |
| RRLDS-NN versus baselines for predicting spike counts | RRLDS-NN · Hippocampus spiking Task #2 | spike count prediction (qualitative) | Improved |
| vary number of latent states (10, 20, 30) (ablation) | RRLDS-NN · Hippocampus Tasks #1 and #2 | estimated rank of dynamics matrix: 5 or 6 for Task #1 and 10 or 11 for Task #2 | No clear effect |
Closest papers by meaning
- Learning Low-dimensional Latent Dynamics from High-dimensional Observations: Non-asymptotics and Lower Bounds (ICML 2024 · similarity 0.89)
- Learning Interpretable Hierarchical Dynamical Systems Models from Time Series Data (ICLR 2025 · similarity 0.88)
- Probabilistic Decomposed Linear Dynamical Systems for Robust Discovery of Latent Neural Dynamics (NeurIPS 2024 · similarity 0.88)
- Modeling Latent Non-Linear Dynamical System over Time Series (AAAI 2025 · similarity 0.87)
- eXponential FAmily Dynamical Systems (XFADS): Large-scale nonlinear Gaussian state-space modeling (NeurIPS 2024 · similarity 0.87)
- Balanced Neural ODEs: nonlinear model order reduction and Koopman operator approximations (ICLR 2025 · similarity 0.87)
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.
