Matches in SemOpenAlex for { <https://semopenalex.org/work/W3103451896> ?p ?o ?g. }
- W3103451896 abstract "How can we design agents that pursue a given objective when all feedback mechanisms are influenceable by the agent? Standard RL algorithms assume a secure reward function, and can thus perform poorly in settings where agents can tamper with the reward-generating mechanism. We present a principled solution to the problem of learning from influenceable feedback, which combines approval with a decoupled feedback collection procedure. For a natural class of corruption functions, decoupled approval algorithms have aligned incentives both at convergence and for their local updates. Empirically, they also scale to complex 3D environments where tampering is possible." @default.
- W3103451896 created "2020-11-23" @default.
- W3103451896 creator A5008987732 @default.
- W3103451896 creator A5019855116 @default.
- W3103451896 creator A5020224050 @default.
- W3103451896 creator A5052300917 @default.
- W3103451896 creator A5059226057 @default.
- W3103451896 creator A5082761102 @default.
- W3103451896 date "2020-11-17" @default.
- W3103451896 modified "2023-09-27" @default.
- W3103451896 title "Avoiding Tampering Incentives in Deep RL via Decoupled Approval." @default.
- W3103451896 cites W145787216 @default.
- W3103451896 cites W1520340982 @default.
- W3103451896 cites W1592847719 @default.
- W3103451896 cites W1624823515 @default.
- W3103451896 cites W1999874108 @default.
- W3103451896 cites W2061562262 @default.
- W3103451896 cites W2099308455 @default.
- W3103451896 cites W2108677974 @default.
- W3103451896 cites W2109910161 @default.
- W3103451896 cites W2110064869 @default.
- W3103451896 cites W2155027007 @default.
- W3103451896 cites W2156869222 @default.
- W3103451896 cites W2165131254 @default.
- W3103451896 cites W2166252986 @default.
- W3103451896 cites W2194690080 @default.
- W3103451896 cites W2312609093 @default.
- W3103451896 cites W2462906003 @default.
- W3103451896 cites W2558634851 @default.
- W3103451896 cites W2604763608 @default.
- W3103451896 cites W2736601468 @default.
- W3103451896 cites W2736629007 @default.
- W3103451896 cites W2770150859 @default.
- W3103451896 cites W2798877128 @default.
- W3103451896 cites W2803281228 @default.
- W3103451896 cites W2896930824 @default.
- W3103451896 cites W2901707424 @default.
- W3103451896 cites W2902711054 @default.
- W3103451896 cites W2908064123 @default.
- W3103451896 cites W2911719076 @default.
- W3103451896 cites W2948625193 @default.
- W3103451896 cites W2950872548 @default.
- W3103451896 cites W2952816888 @default.
- W3103451896 cites W2955240493 @default.
- W3103451896 cites W2962957031 @default.
- W3103451896 cites W2963120839 @default.
- W3103451896 cites W2963277051 @default.
- W3103451896 cites W2963289505 @default.
- W3103451896 cites W2963477884 @default.
- W3103451896 cites W2963489214 @default.
- W3103451896 cites W2963646405 @default.
- W3103451896 cites W2963943581 @default.
- W3103451896 cites W2970011083 @default.
- W3103451896 cites W2970786335 @default.
- W3103451896 cites W2971171756 @default.
- W3103451896 cites W2971870892 @default.
- W3103451896 cites W2995356893 @default.
- W3103451896 cites W3003129838 @default.
- W3103451896 cites W3082042211 @default.
- W3103451896 cites W648152870 @default.
- W3103451896 cites W2770298516 @default.
- W3103451896 hasPublicationYear "2020" @default.
- W3103451896 type Work @default.
- W3103451896 sameAs 3103451896 @default.
- W3103451896 citedByCount "1" @default.
- W3103451896 countsByYear W31034518962019 @default.
- W3103451896 crossrefType "posted-content" @default.
- W3103451896 hasAuthorship W3103451896A5008987732 @default.
- W3103451896 hasAuthorship W3103451896A5019855116 @default.
- W3103451896 hasAuthorship W3103451896A5020224050 @default.
- W3103451896 hasAuthorship W3103451896A5052300917 @default.
- W3103451896 hasAuthorship W3103451896A5059226057 @default.
- W3103451896 hasAuthorship W3103451896A5082761102 @default.
- W3103451896 hasConcept C14036430 @default.
- W3103451896 hasConcept C154945302 @default.
- W3103451896 hasConcept C162324750 @default.
- W3103451896 hasConcept C175444787 @default.
- W3103451896 hasConcept C2777212361 @default.
- W3103451896 hasConcept C2777303404 @default.
- W3103451896 hasConcept C29122968 @default.
- W3103451896 hasConcept C41008148 @default.
- W3103451896 hasConcept C50522688 @default.
- W3103451896 hasConcept C78458016 @default.
- W3103451896 hasConcept C86803240 @default.
- W3103451896 hasConceptScore W3103451896C14036430 @default.
- W3103451896 hasConceptScore W3103451896C154945302 @default.
- W3103451896 hasConceptScore W3103451896C162324750 @default.
- W3103451896 hasConceptScore W3103451896C175444787 @default.
- W3103451896 hasConceptScore W3103451896C2777212361 @default.
- W3103451896 hasConceptScore W3103451896C2777303404 @default.
- W3103451896 hasConceptScore W3103451896C29122968 @default.
- W3103451896 hasConceptScore W3103451896C41008148 @default.
- W3103451896 hasConceptScore W3103451896C50522688 @default.
- W3103451896 hasConceptScore W3103451896C78458016 @default.
- W3103451896 hasConceptScore W3103451896C86803240 @default.
- W3103451896 hasLocation W31034518961 @default.
- W3103451896 hasOpenAccess W3103451896 @default.
- W3103451896 hasPrimaryLocation W31034518961 @default.
- W3103451896 hasRelatedWork W1562694074 @default.
- W3103451896 hasRelatedWork W1998727078 @default.