FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3046419021> ?p ?o ?g. }

• W3046419021 abstract "One of the great promises of robot learning systems is that they will be able to learn from their mistakes and continuously adapt to ever-changing environments. Despite this potential, most of the robot learning systems today are deployed as a fixed policy and they are not being adapted after their deployment. Can we efficiently adapt previously learned behaviors to new environments, objects and percepts in the real world? In this paper, we present a method and empirical evidence towards a robot learning framework that facilitates continuous adaption. In particular, we demonstrate how to adapt vision-based robotic manipulation policies to new variations by fine-tuning via off-policy reinforcement learning, including changes in background, object shape and appearance, lighting conditions, and robot morphology. Further, this adaptation uses less than 0.2% of the data necessary to learn the task from scratch. We find that our approach of adapting pre-trained policies leads to substantial performance gains over the course of fine-tuning, and that pre-training via RL is essential: training from scratch or adapting from supervised ImageNet features are both unsuccessful with such small amounts of data. We also find that these positive results hold in a limited continual learning setting, in which we repeatedly fine-tune a single lineage of policies using data from a succession of new tasks. Our empirical conclusions are consistently supported by experiments on simulated manipulation tasks, and by 52 unique fine-tuning experiments on a real robotic grasping system pre-trained on 580,000 grasps." @default.
• W3046419021 created "2020-08-07" @default.
• W3046419021 creator A5002305939 @default.
• W3046419021 creator A5005431772 @default.
• W3046419021 creator A5026322200 @default.
• W3046419021 creator A5041959064 @default.
• W3046419021 creator A5077367921 @default.
• W3046419021 creator A5088777896 @default.
• W3046419021 date "2020-04-21" @default.
• W3046419021 modified "2023-09-25" @default.
• W3046419021 title "Never Stop Learning: The Effectiveness of Fine-Tuning in Robotic Reinforcement Learning" @default.
• W3046419021 cites W1738827650 @default.
• W3046419021 cites W1977655452 @default.
• W3046419021 cites W1979071892 @default.
• W3046419021 cites W1988071341 @default.
• W3046419021 cites W2097381042 @default.
• W3046419021 cites W2108598243 @default.
• W3046419021 cites W2122838776 @default.
• W3046419021 cites W2122922389 @default.
• W3046419021 cites W2145339207 @default.
• W3046419021 cites W2149933564 @default.
• W3046419021 cites W2153192722 @default.
• W3046419021 cites W2155541015 @default.
• W3046419021 cites W2165698076 @default.
• W3046419021 cites W2167117957 @default.
• W3046419021 cites W2170642268 @default.
• W3046419021 cites W2194775991 @default.
• W3046419021 cites W2201912979 @default.
• W3046419021 cites W2510153535 @default.
• W3046419021 cites W2528489519 @default.
• W3046419021 cites W2534269850 @default.
• W3046419021 cites W2583137229 @default.
• W3046419021 cites W2592538810 @default.
• W3046419021 cites W2605102758 @default.
• W3046419021 cites W2606327391 @default.
• W3046419021 cites W2624871570 @default.
• W3046419021 cites W2737821837 @default.
• W3046419021 cites W2755546070 @default.
• W3046419021 cites W2784121710 @default.
• W3046419021 cites W2804935296 @default.
• W3046419021 cites W2811024793 @default.
• W3046419021 cites W2886380958 @default.
• W3046419021 cites W2887280559 @default.
• W3046419021 cites W2890208753 @default.
• W3046419021 cites W2895558617 @default.
• W3046419021 cites W2897170587 @default.
• W3046419021 cites W2908470496 @default.
• W3046419021 cites W2914584948 @default.
• W3046419021 cites W2918049070 @default.
• W3046419021 cites W2925173345 @default.
• W3046419021 cites W2952791429 @default.
• W3046419021 cites W2962793652 @default.
• W3046419021 cites W2962812027 @default.
• W3046419021 cites W2963026768 @default.
• W3046419021 cites W2963276406 @default.
• W3046419021 cites W2963341956 @default.
• W3046419021 cites W2963634205 @default.
• W3046419021 cites W2964021598 @default.
• W3046419021 cites W2964093801 @default.
• W3046419021 cites W2964112890 @default.
• W3046419021 cites W2964118020 @default.
• W3046419021 cites W2964161785 @default.
• W3046419021 cites W2964342357 @default.
• W3046419021 cites W2975909688 @default.
• W3046419021 cites W2980820015 @default.
• W3046419021 cites W2981030070 @default.
• W3046419021 cites W2984165007 @default.
• W3046419021 cites W2988603490 @default.
• W3046419021 cites W2988640752 @default.
• W3046419021 cites W2990216309 @default.
• W3046419021 cites W3001500946 @default.
• W3046419021 cites W3004116079 @default.
• W3046419021 cites W3028676366 @default.
• W3046419021 cites W3034909023 @default.
• W3046419021 cites W3101442004 @default.
• W3046419021 cites W3130717831 @default.
• W3046419021 hasPublicationYear "2020" @default.
• W3046419021 type Work @default.
• W3046419021 sameAs 3046419021 @default.
• W3046419021 citedByCount "11" @default.
• W3046419021 countsByYear W30464190212020 @default.
• W3046419021 countsByYear W30464190212021 @default.
• W3046419021 countsByYear W30464190212022 @default.
• W3046419021 crossrefType "posted-content" @default.
• W3046419021 hasAuthorship W3046419021A5002305939 @default.
• W3046419021 hasAuthorship W3046419021A5005431772 @default.
• W3046419021 hasAuthorship W3046419021A5026322200 @default.
• W3046419021 hasAuthorship W3046419021A5041959064 @default.
• W3046419021 hasAuthorship W3046419021A5077367921 @default.
• W3046419021 hasAuthorship W3046419021A5088777896 @default.
• W3046419021 hasConcept C105339364 @default.
• W3046419021 hasConcept C107457646 @default.
• W3046419021 hasConcept C111919701 @default.
• W3046419021 hasConcept C120665830 @default.
• W3046419021 hasConcept C121332964 @default.
• W3046419021 hasConcept C127413603 @default.
• W3046419021 hasConcept C139807058 @default.
• W3046419021 hasConcept C154945302 @default.
• W3046419021 hasConcept C188888258 @default.
• W3046419021 hasConcept C19966478 @default.

• next