FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3036670859> ?p ?o ?g. }

• W3036670859 abstract "Deep reinforcement learning (RL) agents often fail to generalize to unseen scenarios, even when they are trained on many instances of semantically similar environments. Data augmentation has recently been shown to improve the sample efficiency and generalization of RL agents. However, different tasks tend to benefit from different kinds of data augmentation. In this paper, we compare three approaches for automatically finding an appropriate augmentation. These are combined with two novel regularization terms for the policy and value function, required to make the use of data augmentation theoretically sound for certain actor-critic algorithms. We evaluate our methods on the Procgen benchmark which consists of 16 procedurally-generated environments and show that it improves test performance by ~40% relative to standard RL algorithms. Our agent outperforms other baselines specifically designed to improve generalization in RL. In addition, we show that our agent learns policies and representations that are more robust to changes in the environment that do not affect the agent, such as the background. Our implementation is available at this https URL." @default.
• W3036670859 created "2020-06-25" @default.
• W3036670859 creator A5000725688 @default.
• W3036670859 creator A5018702533 @default.
• W3036670859 creator A5071892594 @default.
• W3036670859 creator A5086657309 @default.
• W3036670859 creator A5089960673 @default.
• W3036670859 date "2020-06-23" @default.
• W3036670859 modified "2023-09-27" @default.
• W3036670859 title "Automatic Data Augmentation for Generalization in Deep Reinforcement Learning." @default.
• W3036670859 cites W1757796397 @default.
• W3036670859 cites W1771410628 @default.
• W3036670859 cites W2063971957 @default.
• W3036670859 cites W2064675550 @default.
• W3036670859 cites W2076337359 @default.
• W3036670859 cites W2095705004 @default.
• W3036670859 cites W2097356275 @default.
• W3036670859 cites W2112796928 @default.
• W3036670859 cites W2119717200 @default.
• W3036670859 cites W2141125852 @default.
• W3036670859 cites W2144796873 @default.
• W3036670859 cites W2147800946 @default.
• W3036670859 cites W2156163116 @default.
• W3036670859 cites W2550182557 @default.
• W3036670859 cites W2556958149 @default.
• W3036670859 cites W2605102758 @default.
• W3036670859 cites W2619947201 @default.
• W3036670859 cites W2736601468 @default.
• W3036670859 cites W2746314669 @default.
• W3036670859 cites W2781585732 @default.
• W3036670859 cites W2786036274 @default.
• W3036670859 cites W2796979132 @default.
• W3036670859 cites W2809668646 @default.
• W3036670859 cites W2891790128 @default.
• W3036670859 cites W2893662673 @default.
• W3036670859 cites W2898436992 @default.
• W3036670859 cites W2903181768 @default.
• W3036670859 cites W2949117887 @default.
• W3036670859 cites W2949608212 @default.
• W3036670859 cites W2949694312 @default.
• W3036670859 cites W2949736877 @default.
• W3036670859 cites W2951775809 @default.
• W3036670859 cites W2962715211 @default.
• W3036670859 cites W2962749646 @default.
• W3036670859 cites W2963403143 @default.
• W3036670859 cites W2963913081 @default.
• W3036670859 cites W2964121744 @default.
• W3036670859 cites W2970214542 @default.
• W3036670859 cites W2978409868 @default.
• W3036670859 cites W2979579363 @default.
• W3036670859 cites W2980370789 @default.
• W3036670859 cites W2987283559 @default.
• W3036670859 cites W2994073215 @default.
• W3036670859 cites W2994536315 @default.
• W3036670859 cites W2996110979 @default.
• W3036670859 cites W2996283175 @default.
• W3036670859 cites W2999617596 @default.
• W3036670859 cites W3002447977 @default.
• W3036670859 cites W3005680577 @default.
• W3036670859 cites W3011651653 @default.
• W3036670859 cites W3021475836 @default.
• W3036670859 cites W3021708257 @default.
• W3036670859 cites W3023640063 @default.
• W3036670859 cites W3029947299 @default.
• W3036670859 cites W3033384842 @default.
• W3036670859 cites W3035595221 @default.
• W3036670859 cites W3036185205 @default.
• W3036670859 cites W3037211759 @default.
• W3036670859 cites W3037871539 @default.
• W3036670859 cites W3085605093 @default.
• W3036670859 cites W3092744199 @default.
• W3036670859 cites W3094044532 @default.
• W3036670859 cites W3097975205 @default.
• W3036670859 cites W3115293622 @default.
• W3036670859 cites W3119486431 @default.
• W3036670859 cites W3092271956 @default.
• W3036670859 hasPublicationYear "2020" @default.
• W3036670859 type Work @default.
• W3036670859 sameAs 3036670859 @default.
• W3036670859 citedByCount "23" @default.
• W3036670859 countsByYear W30366708592020 @default.
• W3036670859 countsByYear W30366708592021 @default.
• W3036670859 crossrefType "posted-content" @default.
• W3036670859 hasAuthorship W3036670859A5000725688 @default.
• W3036670859 hasAuthorship W3036670859A5018702533 @default.
• W3036670859 hasAuthorship W3036670859A5071892594 @default.
• W3036670859 hasAuthorship W3036670859A5086657309 @default.
• W3036670859 hasAuthorship W3036670859A5089960673 @default.
• W3036670859 hasConcept C119857082 @default.
• W3036670859 hasConcept C13280743 @default.
• W3036670859 hasConcept C134306372 @default.
• W3036670859 hasConcept C14036430 @default.
• W3036670859 hasConcept C154945302 @default.
• W3036670859 hasConcept C177148314 @default.
• W3036670859 hasConcept C185798385 @default.
• W3036670859 hasConcept C205649164 @default.
• W3036670859 hasConcept C2776135515 @default.
• W3036670859 hasConcept C33923547 @default.
• W3036670859 hasConcept C41008148 @default.
• W3036670859 hasConcept C78458016 @default.

• next