2021年最新理工科硕士研究生毕业答辩ppt-08.pptxVIP

Master Thesis ProposalImproving the Deep Reinforcement Learning Training Efficiency in High-Dimensional SapceSupervisor：Guanfeng Liu、Juan Rojas Presenter：Yijiong Lin2018.12.28OutlineBackgrounds Relative WorksCritical IssuesMethodology ExperimentExpected ContributionsProgress Anticipated AchievementsBackgroundsPolicyExperience-driven Autonomous LearningThe goal of reinforcement learningWhat DRL can do?End-to-End Training of Deep Visuomotor Policies Abeel et. al, 2015Demonstration Augmented Policy GradientRajeswaran et. al, 2018HammerPen OrientationOpen DoorRelocationQT-Opt: Scalable DRL for V