Research map / Time series / Causal and generative time series
Dynamical Wasserstein Barycenters for Time-series Modeling
Kevin C Cheng, Shuchin Aeron, Michael C Hughes, Eric Miller
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 |
|---|---|---|---|
| use Wasserstein barycentric interpolation instead of linear Gaussian-mixture interpolation | DWB · MSR | e_nll: 1.50 -> 1.02 | Improved |
| use Wasserstein barycentric interpolation instead of linear Gaussian-mixture interpolation | DWB · MSR | e_W | Improved |
| use DWB rather than a comparable-parameter DSS model | MSR · MSR | e_W: 4.34 -> 0.27 | Improved |
| use DWB rather than a comparable-parameter DSS model | MSR · MSR | e_nll: 1.49 -> 1.02 | Improved |
| use DWB rather than a much larger DSS model | MSR · MSR | e_W: 3.07 -> 0.27 | Improved |
| use a much larger DSS model rather than DWB | MSR · MSR | e_nll: 1.02 -> -0.508 | Improved |
| replace the learnable two-component Beta mixture with a single fixed Beta prior (ablation) | DWB · MSR | e_W: 0.27 -> 0.50 | Hurt |
| optimize Gaussian emission parameters with Wasserstein-Riemannian geometry instead of Euclidean geometry | DWB · simulated time series | optimization convergence speed | Improved |
| optimize Gaussian emission parameters with Wasserstein-Riemannian geometry instead of Euclidean geometry | DWB · simulated time series | converged solution | No clear effect |
| increase the data-fit weight λ | DWB · MSR | model fit | Improved |
| initialize pure-state parameters with ground-truth labels rather than unsupervised clustering | DWB · MSR | e_W: 0.011 and 0.022 | No clear effect |
Closest papers by meaning
- Learning Interpretable Hierarchical Dynamical Systems Models from Time Series Data (ICLR 2025 · similarity 0.84)
- Identifying latent state transitions in non-linear dynamical systems (ICLR 2025 · similarity 0.84)
- Time series analysis with Gumbel dynamics (NeurIPS 2026 · similarity 0.84)
- A scalable generative model for dynamical system reconstruction from neuroimaging data (NeurIPS 2024 · similarity 0.84)
- A Unified Approach for Computing Wasserstein Barycenters of Discrete and Continuous Measures (NeurIPS 2026 · similarity 0.84)
- Fréchet Regression on the Bures-Wasserstein Manifold (NeurIPS 2026 · 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.
