Research map / Time series / Causal and generative time series
Discriminative State Space Models
Vitaly Kuznetsov, Mehryar Mohri
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 |
|---|---|---|---|
| Learn state predictor jointly with observation predictor via SRM over discrepancy-based bound (Theorem 2/3) instead of fixing state-space sequence | Discriminative State-Space Models · theoretical analysis (non-stationary time series) | generalization bound | No clear effect |
| Use state-space discrepancy disc(s) instead of sup over state predictors of disc(s) in the bound | DSSM generalization bound · theoretical example with deterministic periodic/constant dynamics | bound tightness | Improved |
| Optimistic SRM objective omitting the discrepancy penalty (under stability Assumption 1) | SRM for DSSMs · theoretical analysis | generalization bound | Improved |
| Boosted structural time series model (BOOSTSM) with convex objective via coordinate descent, adding trees or state trajectories | BOOSTSM · preliminary experiments (Appendix D, not reported in main text) | forecasting accuracy | No clear effect |
Closest papers by meaning
- Probabilistic Recurrent State-Space Models (ICML 2018 · similarity 0.86)
- From Generalization Analysis to Optimization Designs for State Space Models (ICML 2024 · similarity 0.86)
- Continuous Domain Generalization (NeurIPS 2025 · similarity 0.84)
- Structured Matrix Basis for Multivariate Time Series Forecasting with Interpretable Dynamics (NeurIPS 2024 · similarity 0.83)
- Learning invariant representations of time-homogeneous stochastic dynamical systems (ICLR 2024 · similarity 0.83)
- Probabilistic Forecasting with Stochastic Interpolants and Föllmer Processes (ICML 2024 · 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.
