Matches in SemOpenAlex for { <https://semopenalex.org/work/W3100803008> ?p ?o ?g. }
- W3100803008 abstract "When designing controllers for safety-critical systems, practitioners often face a challenging tradeoff between robustness and performance. While robust control methods provide rigorous guarantees on system stability under certain worst-case disturbances, they often yield simple controllers that perform poorly in the average (non-worst) case. In contrast, nonlinear control methods trained using deep learning have achieved state-of-the-art performance on many control tasks, but often lack robustness guarantees. In this paper, we propose a technique that combines the strengths of these two approaches: constructing a generic nonlinear control policy class, parameterized by neural networks, that nonetheless enforces the same provable robustness criteria as robust control. Specifically, our approach entails integrating custom convex-optimization-based projection layers into a neural network-based policy. We demonstrate the power of this approach on several domains, improving in average-case performance over existing robust control methods and in worst-case stability over (non-robust) deep RL methods." @default.
- W3100803008 created "2020-11-23" @default.
- W3100803008 creator A5043154223 @default.
- W3100803008 creator A5058043823 @default.
- W3100803008 creator A5075035644 @default.
- W3100803008 creator A5075620331 @default.
- W3100803008 date "2020-11-16" @default.
- W3100803008 modified "2023-09-27" @default.
- W3100803008 title "Enforcing robust control guarantees within neural network policies" @default.
- W3100803008 cites W1518931405 @default.
- W3100803008 cites W1536997931 @default.
- W3100803008 cites W1557931734 @default.
- W3100803008 cites W1579118835 @default.
- W3100803008 cites W1976410223 @default.
- W3100803008 cites W1990401127 @default.
- W3100803008 cites W2026932193 @default.
- W3100803008 cites W2052399669 @default.
- W3100803008 cites W2053572490 @default.
- W3100803008 cites W2055891887 @default.
- W3100803008 cites W2105078254 @default.
- W3100803008 cites W2116479681 @default.
- W3100803008 cites W2145339207 @default.
- W3100803008 cites W2171224297 @default.
- W3100803008 cites W2296319761 @default.
- W3100803008 cites W2413518943 @default.
- W3100803008 cites W2557925931 @default.
- W3100803008 cites W2602963933 @default.
- W3100803008 cites W2618318883 @default.
- W3100803008 cites W2736601468 @default.
- W3100803008 cites W2737906791 @default.
- W3100803008 cites W2751681791 @default.
- W3100803008 cites W2788084076 @default.
- W3100803008 cites W2897661175 @default.
- W3100803008 cites W2898291697 @default.
- W3100803008 cites W2914474570 @default.
- W3100803008 cites W2950321944 @default.
- W3100803008 cites W2962803570 @default.
- W3100803008 cites W2963958573 @default.
- W3100803008 cites W2963970238 @default.
- W3100803008 cites W2964346299 @default.
- W3100803008 cites W2970791107 @default.
- W3100803008 cites W2970795963 @default.
- W3100803008 cites W2981030070 @default.
- W3100803008 cites W2982428006 @default.
- W3100803008 cites W2989800586 @default.
- W3100803008 cites W2994712737 @default.
- W3100803008 cites W3082066778 @default.
- W3100803008 cites W3141595720 @default.
- W3100803008 cites W645921674 @default.
- W3100803008 hasPublicationYear "2020" @default.
- W3100803008 type Work @default.
- W3100803008 sameAs 3100803008 @default.
- W3100803008 citedByCount "7" @default.
- W3100803008 countsByYear W31008030082020 @default.
- W3100803008 countsByYear W31008030082021 @default.
- W3100803008 countsByYear W31008030082022 @default.
- W3100803008 crossrefType "posted-content" @default.
- W3100803008 hasAuthorship W3100803008A5043154223 @default.
- W3100803008 hasAuthorship W3100803008A5058043823 @default.
- W3100803008 hasAuthorship W3100803008A5075035644 @default.
- W3100803008 hasAuthorship W3100803008A5075620331 @default.
- W3100803008 hasConcept C104317684 @default.
- W3100803008 hasConcept C11413529 @default.
- W3100803008 hasConcept C121332964 @default.
- W3100803008 hasConcept C126255220 @default.
- W3100803008 hasConcept C154945302 @default.
- W3100803008 hasConcept C158622935 @default.
- W3100803008 hasConcept C165464430 @default.
- W3100803008 hasConcept C185592680 @default.
- W3100803008 hasConcept C2775924081 @default.
- W3100803008 hasConcept C31531917 @default.
- W3100803008 hasConcept C33923547 @default.
- W3100803008 hasConcept C41008148 @default.
- W3100803008 hasConcept C47446073 @default.
- W3100803008 hasConcept C50644808 @default.
- W3100803008 hasConcept C55493867 @default.
- W3100803008 hasConcept C62520636 @default.
- W3100803008 hasConcept C63479239 @default.
- W3100803008 hasConceptScore W3100803008C104317684 @default.
- W3100803008 hasConceptScore W3100803008C11413529 @default.
- W3100803008 hasConceptScore W3100803008C121332964 @default.
- W3100803008 hasConceptScore W3100803008C126255220 @default.
- W3100803008 hasConceptScore W3100803008C154945302 @default.
- W3100803008 hasConceptScore W3100803008C158622935 @default.
- W3100803008 hasConceptScore W3100803008C165464430 @default.
- W3100803008 hasConceptScore W3100803008C185592680 @default.
- W3100803008 hasConceptScore W3100803008C2775924081 @default.
- W3100803008 hasConceptScore W3100803008C31531917 @default.
- W3100803008 hasConceptScore W3100803008C33923547 @default.
- W3100803008 hasConceptScore W3100803008C41008148 @default.
- W3100803008 hasConceptScore W3100803008C47446073 @default.
- W3100803008 hasConceptScore W3100803008C50644808 @default.
- W3100803008 hasConceptScore W3100803008C55493867 @default.
- W3100803008 hasConceptScore W3100803008C62520636 @default.
- W3100803008 hasConceptScore W3100803008C63479239 @default.
- W3100803008 hasLocation W31008030081 @default.
- W3100803008 hasOpenAccess W3100803008 @default.
- W3100803008 hasPrimaryLocation W31008030081 @default.
- W3100803008 hasRelatedWork W163289901 @default.
- W3100803008 hasRelatedWork W2003601090 @default.