Research map / Probabilistic and scientific modeling / Dynamical systems
Learning Koopman Representations with Controllability Guarantees
Keyan Miao, Han Wang, Xuda Ding, Konstantinos Gatsis, Andreas Krause, Antonis Papachristodoulou
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 |
|---|---|---|---|
| enforce controllability via canonical parameterization of Koopman matrices | DKO · Mountain Car (1% data) | prediction MSE: 0.0200 -> 0.0032 | Improved |
| enforce controllability | DKO · Pendulum (1% data) | prediction MSE: 1.5347 -> 0.3747 | Improved |
| enforce controllability | DKO · Cartpole (1% data) | prediction MSE: 0.1306 -> 0.0095 | Improved |
| enforce controllability with abundant data (50-100%) | DKO · Mountain Car / Pendulum / Cartpole | prediction MSE | No clear effect |
| Koopman controllable model instead of plain MLP at 1% data | MLP · Pendulum (1% data) | prediction MSE: 1.1778 -> 0.3747 | Improved |
| use controllable model in MPC with 5% data | DKO · Mountain Car (5% data) | MPC input cost: 297.51 -> 165.21 | Improved |
| use controllable model in MPC with 10% data | DKO · Pendulum swing-up (10% data) | MPC input cost: 239.2 -> 43.50 | Improved |
| use controllable model in MPC with scarce data | DKO · Mountain Car 1%, Pendulum 5%, Cartpole 30% | MPC task completion | Improved |
| controllable model on multi-input GRN | DKO · GRN 6-dim | test MSE: 3.0 × 10−4 -> 1.0 × 10−4 | Improved |
| controllable model on GRN control | DKO · GRN setpoint 6 | MPC input cost: 89.98 -> 80.96 | Improved |
| enforce controllability prior | DKO · Mountain Car, Pendulum, Cartpole | training convergence | Improved |
Closest papers by meaning
- Deep-Koopman-KANDy: Dictionary Discovery for Deep-Koopman Operators with Kolmogorov-Arnold Networks for Dynamics (NeurIPS 2026 · similarity 0.87)
- Course Correcting Koopman Representations (ICLR 2024 · similarity 0.86)
- ResKoopNet: Learning Koopman Representations for Complex Dynamics with Spectral Residuals (ICML 2025 · similarity 0.86)
- Learning Hybrid Dynamics Models with Simulator-Informed Latent States (AAAI 2024 · similarity 0.85)
- Identifying latent state transitions in non-linear dynamical systems (ICLR 2025 · similarity 0.85)
- Efficient Dynamics Modeling in Interactive Environments with Koopman Theory (ICLR 2024 · similarity 0.85)
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.
