Research map / Probabilistic and scientific modeling / Dynamical systems
OVERT: An Algorithm for Safety Verification of Neural Network Control Policies for Nonlinear Systems
Chelsea Sidrane, Amir Maleki, Ahmed Irfan, Mykel J. Kochenderfer
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 OVERT reachable set computation instead of NNV | closed-loop verification · S1 single pendulum | runtime (s) and property proof: 483 -> 99 | Improved |
| use OVERT reachable set computation instead of NNV | closed-loop verification · T1 TORA | runtime: Time Out -> 232 | Improved |
| use OVERT feasibility framing instead of dReal | closed-loop verification · S1 single pendulum | runtime (s): Time Out -> 77 | Improved |
| use OVERT feasibility framing instead of dReal | closed-loop verification · T1 TORA | runtime (s): 12655 -> 79 | Improved |
| use feasibility framing instead of reachable-set framing | OVERT · S1 single pendulum | runtime (s): 99 -> 77 | Improved |
| use feasibility framing instead of reachable-set framing | OVERT · T1 TORA | runtime (s): 232 -> 79 | Improved |
| use hybrid symbolic queries instead of naive one-step concrete sets (ablation) | OVERT reachability · S1/S2 single pendulum | safety property proven | Improved |
| use hybrid symbolic queries instead of naive one-step concrete sets (ablation) | OVERT reachability · S2 single pendulum | runtime (minutes): less than 4 minutes -> just under 12 minutes | Hurt |
| use hybrid symbolic queries instead of concrete sets (ablation) | OVERT reachability · Adaptive cruise control | reachable set tightness | Mixed |
| increase number of linear segments n in the overapproximation (ablation) | OVERT MIP · benchmarks | MIP solve difficulty | No clear effect |
| increase neural network size (T1 to T3 TORA) (ablation) | OVERT reach · TORA | runtime | Hurt |
Closest papers by meaning
- Verified Safe Reinforcement Learning for Neural Network Dynamic Models (NeurIPS 2024 · similarity 0.89)
- Neural Abstractions (NeurIPS 2022 · similarity 0.89)
- Provably Safe Neural Network Controllers via Differential Dynamic Logic (NeurIPS 2024 · similarity 0.88)
- SEEV: Synthesis with Efficient Exact Verification for ReLU Neural Barrier Functions (NeurIPS 2024 · similarity 0.87)
- Lyapunov-stable Neural Control for State and Output Feedback: A Novel Formulation (ICML 2024 · similarity 0.86)
- Stability Analysis of Switched Linear Systems with Neural Lyapunov Functions (AAAI 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.
