Matches in SemOpenAlex for { <https://semopenalex.org/work/W4200436615> ?p ?o ?g. }
Showing items 1 to 75 of
75
with 100 items per page.
- W4200436615 abstract "In edge computing it is pivotal to automatically manage the resources to increase efficiency with limited resources. Deep reinforcement learning, which aims to maximize the long term cumulative reward, has recently been adopted in such scenarios. However, training the policy in real-world edge computing environment is challenging since arbitrary exploration in real world could drastically impair user utilities. In this paper, we propose a novel imitation learning approach to construct a virtual environment, in which the policy can be trained freely without additional costs. Under the virtual environment, we use a multi-agent reinforcement learning to manage the edge resources. Our method adopts a decentralized, sequential approach to deal with the uncertainties in the environment. Specifically, we decompose the target reward function into separated global part and local parts, where the global part is shared by all agents for cooperation, and the local parts are owned by each individual edge to accelerate the model learning. Extensive experimental results demonstrate that the constructed environment is very close to the real environment. In addition, the proposed multi agent reinforcement learning algorithm can converge very fast in the training phase and outperforms other state-of-art methods significantly in a variety of scenarios." @default.
- W4200436615 created "2021-12-31" @default.
- W4200436615 creator A5002899652 @default.
- W4200436615 creator A5010092389 @default.
- W4200436615 creator A5010400154 @default.
- W4200436615 creator A5019908722 @default.
- W4200436615 creator A5036996300 @default.
- W4200436615 creator A5052643678 @default.
- W4200436615 creator A5066716873 @default.
- W4200436615 creator A5075218726 @default.
- W4200436615 date "2021-09-01" @default.
- W4200436615 modified "2023-09-24" @default.
- W4200436615 title "Multi-Agent Reinforcement Learning for Edge Resource Management with Reconstructed Environment" @default.
- W4200436615 doi "https://doi.org/10.1109/ispa-bdcloud-socialcom-sustaincom52081.2021.00233" @default.
- W4200436615 hasPublicationYear "2021" @default.
- W4200436615 type Work @default.
- W4200436615 citedByCount "0" @default.
- W4200436615 crossrefType "proceedings-article" @default.
- W4200436615 hasAuthorship W4200436615A5002899652 @default.
- W4200436615 hasAuthorship W4200436615A5010092389 @default.
- W4200436615 hasAuthorship W4200436615A5010400154 @default.
- W4200436615 hasAuthorship W4200436615A5019908722 @default.
- W4200436615 hasAuthorship W4200436615A5036996300 @default.
- W4200436615 hasAuthorship W4200436615A5052643678 @default.
- W4200436615 hasAuthorship W4200436615A5066716873 @default.
- W4200436615 hasAuthorship W4200436615A5075218726 @default.
- W4200436615 hasConcept C107457646 @default.
- W4200436615 hasConcept C111919701 @default.
- W4200436615 hasConcept C120314980 @default.
- W4200436615 hasConcept C126388530 @default.
- W4200436615 hasConcept C136197465 @default.
- W4200436615 hasConcept C14036430 @default.
- W4200436615 hasConcept C154945302 @default.
- W4200436615 hasConcept C15744967 @default.
- W4200436615 hasConcept C162307627 @default.
- W4200436615 hasConcept C25344961 @default.
- W4200436615 hasConcept C2778456923 @default.
- W4200436615 hasConcept C41008148 @default.
- W4200436615 hasConcept C77805123 @default.
- W4200436615 hasConcept C78458016 @default.
- W4200436615 hasConcept C86803240 @default.
- W4200436615 hasConcept C97541855 @default.
- W4200436615 hasConceptScore W4200436615C107457646 @default.
- W4200436615 hasConceptScore W4200436615C111919701 @default.
- W4200436615 hasConceptScore W4200436615C120314980 @default.
- W4200436615 hasConceptScore W4200436615C126388530 @default.
- W4200436615 hasConceptScore W4200436615C136197465 @default.
- W4200436615 hasConceptScore W4200436615C14036430 @default.
- W4200436615 hasConceptScore W4200436615C154945302 @default.
- W4200436615 hasConceptScore W4200436615C15744967 @default.
- W4200436615 hasConceptScore W4200436615C162307627 @default.
- W4200436615 hasConceptScore W4200436615C25344961 @default.
- W4200436615 hasConceptScore W4200436615C2778456923 @default.
- W4200436615 hasConceptScore W4200436615C41008148 @default.
- W4200436615 hasConceptScore W4200436615C77805123 @default.
- W4200436615 hasConceptScore W4200436615C78458016 @default.
- W4200436615 hasConceptScore W4200436615C86803240 @default.
- W4200436615 hasConceptScore W4200436615C97541855 @default.
- W4200436615 hasFunder F4320336329 @default.
- W4200436615 hasLocation W42004366151 @default.
- W4200436615 hasOpenAccess W4200436615 @default.
- W4200436615 hasPrimaryLocation W42004366151 @default.
- W4200436615 hasRelatedWork W2776387983 @default.
- W4200436615 hasRelatedWork W2795910581 @default.
- W4200436615 hasRelatedWork W2900070427 @default.
- W4200436615 hasRelatedWork W3013056451 @default.
- W4200436615 hasRelatedWork W3037422413 @default.
- W4200436615 hasRelatedWork W3089192431 @default.
- W4200436615 hasRelatedWork W3111983280 @default.
- W4200436615 hasRelatedWork W3208790065 @default.
- W4200436615 hasRelatedWork W4309771267 @default.
- W4200436615 hasRelatedWork W4312815336 @default.
- W4200436615 isParatext "false" @default.
- W4200436615 isRetracted "false" @default.
- W4200436615 workType "article" @default.