FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4384026515> ?p ?o ?g. }

• W4384026515 abstract "Deep reinforcement learning (DRL) has become a dominant paradigm for using deep learning to carry out tasks where complex policies are learned for reactive systems. However, these policies are “black-boxes”, e.g., opaque to humans and known to be susceptible to bugs. For example, it is hard - if not impossible - to guarantee that the trained DRL agent adheres to specific safety and fairness properties that may be required. This doctoral dissertation's first and primary contribution is a novel approach to developing DRL agents, which will improve the DRL training process by pushing the learned policy toward high performance on its main task and compliance with such safety and fairness properties, guaranteeing a high probability of compliance while not compromising the performance of the resulting agent. The approach is realized by incorporating domain-specific knowledge captured as key properties defined by domain experts directly into the DRL optimization process while leveraging behavioral languages that are natural to the domain experts. We have validated the proposed approach by extending the AI-Gym Python framework [1] for training DRL agents and integrating it with the BP-Py framework [2] for specifying scenario-based models [3] in a way that allows scenario objects to affect the training process through reward and cost functions, demonstrating dramatic improvement in the safety and performance of the agent. In addition, we have validated the resulting DRL agents using the Marabou verifier [4], confirming that the resulting agents indeed comply (in full) with the required safety and fairness properties. We have applied the approach, training DRL agents for use cases from network communication and robotic navigation domains, exhibiting strong results. A second contribution of this doctoral dissertation is to develop and leverage probabilistic verification methods for deep neural networks to overcome the current scalability limitations of neural network verification technology, limiting the applicability of verification to practical DRL agents. We carried out an initial validation of the concept in the domain of image classification, showing promising results." @default.
• W4384026515 created "2023-07-13" @default.
• W4384026515 creator A5058668708 @default.
• W4384026515 date "2023-05-01" @default.
• W4384026515 modified "2023-09-23" @default.
• W4384026515 title "Enhancing Deep Reinforcement Learning with Executable Specifications" @default.
• W4384026515 cites W2075140220 @default.
• W4384026515 cites W2099529102 @default.
• W4384026515 cites W2339341469 @default.
• W4384026515 cites W2957311447 @default.
• W4384026515 cites W2996896271 @default.
• W4384026515 cites W3039941973 @default.
• W4384026515 cites W3044557985 @default.
• W4384026515 cites W3139593476 @default.
• W4384026515 cites W3150718622 @default.
• W4384026515 cites W3188209232 @default.
• W4384026515 cites W3209627429 @default.
• W4384026515 cites W4308080966 @default.
• W4384026515 cites W4315646410 @default.
• W4384026515 cites W4366723927 @default.
• W4384026515 cites W3021248082 @default.
• W4384026515 doi "https://doi.org/10.1109/icse-companion58688.2023.00058" @default.
• W4384026515 hasPublicationYear "2023" @default.
• W4384026515 type Work @default.
• W4384026515 citedByCount "0" @default.
• W4384026515 crossrefType "proceedings-article" @default.
• W4384026515 hasAuthorship W4384026515A5058668708 @default.
• W4384026515 hasConcept C107457646 @default.
• W4384026515 hasConcept C115903868 @default.
• W4384026515 hasConcept C119857082 @default.
• W4384026515 hasConcept C127413603 @default.
• W4384026515 hasConcept C134306372 @default.
• W4384026515 hasConcept C154945302 @default.
• W4384026515 hasConcept C160145156 @default.
• W4384026515 hasConcept C199360897 @default.
• W4384026515 hasConcept C201995342 @default.
• W4384026515 hasConcept C2780451532 @default.
• W4384026515 hasConcept C33923547 @default.
• W4384026515 hasConcept C36503486 @default.
• W4384026515 hasConcept C41008148 @default.
• W4384026515 hasConcept C519991488 @default.
• W4384026515 hasConcept C97541855 @default.
• W4384026515 hasConcept C98045186 @default.
• W4384026515 hasConceptScore W4384026515C107457646 @default.
• W4384026515 hasConceptScore W4384026515C115903868 @default.
• W4384026515 hasConceptScore W4384026515C119857082 @default.
• W4384026515 hasConceptScore W4384026515C127413603 @default.
• W4384026515 hasConceptScore W4384026515C134306372 @default.
• W4384026515 hasConceptScore W4384026515C154945302 @default.
• W4384026515 hasConceptScore W4384026515C160145156 @default.
• W4384026515 hasConceptScore W4384026515C199360897 @default.
• W4384026515 hasConceptScore W4384026515C201995342 @default.
• W4384026515 hasConceptScore W4384026515C2780451532 @default.
• W4384026515 hasConceptScore W4384026515C33923547 @default.
• W4384026515 hasConceptScore W4384026515C36503486 @default.
• W4384026515 hasConceptScore W4384026515C41008148 @default.
• W4384026515 hasConceptScore W4384026515C519991488 @default.
• W4384026515 hasConceptScore W4384026515C97541855 @default.
• W4384026515 hasConceptScore W4384026515C98045186 @default.
• W4384026515 hasLocation W43840265151 @default.
• W4384026515 hasOpenAccess W4384026515 @default.
• W4384026515 hasPrimaryLocation W43840265151 @default.
• W4384026515 hasRelatedWork W1589273879 @default.
• W4384026515 hasRelatedWork W2774891019 @default.
• W4384026515 hasRelatedWork W2891993883 @default.
• W4384026515 hasRelatedWork W2979801952 @default.
• W4384026515 hasRelatedWork W3022038857 @default.
• W4384026515 hasRelatedWork W4285815787 @default.
• W4384026515 hasRelatedWork W4297816585 @default.
• W4384026515 hasRelatedWork W4312949351 @default.
• W4384026515 hasRelatedWork W4319083788 @default.
• W4384026515 hasRelatedWork W4320008018 @default.
• W4384026515 isParatext "false" @default.
• W4384026515 isRetracted "false" @default.
• W4384026515 workType "article" @default.