Matches in SemOpenAlex for { <https://semopenalex.org/work/W4312910724> ?p ?o ?g. }
Showing items 1 to 80 of
80
with 100 items per page.
- W4312910724 endingPage "184" @default.
- W4312910724 startingPage "173" @default.
- W4312910724 abstract "By learning policy directly from high-dimensional visual inputs, e.g., video frames, Deep Reinforcement Learning (DRL) has achieved great successes for solving sequential decision-making problems, where 2D convolutional network is usually adopted for extracting the underlying spatial features. However, such spatial feature extraction methods do not consider the temporal information existed in the input frames. To address this issue, Transformer, 3D convolutional network and Long Short-Term Memory (LSTM) have been used in DRL, but often result in excessive increase of model parameters and computation cost. Furthermore, multiple down-sampling of images will lead to the loss of sequential information. In this paper, we propose a novel model for extracting sequential information, namely temporal aggregation network (TAN). Comparing with existing methods, TAN can extract the sequential information without the needs of multiple downsampling of consecutive images. Moreover, by decoupling the computation between spatial and channel dimensions, a lightweight model is built in TAN. Experiments in classic Atari 2600 games show that our method can improve the efficiency of decision-making of DRL algorithms compared with baselines." @default.
- W4312910724 created "2023-01-05" @default.
- W4312910724 creator A5011793745 @default.
- W4312910724 creator A5031014820 @default.
- W4312910724 creator A5051623896 @default.
- W4312910724 date "2022-01-01" @default.
- W4312910724 modified "2023-10-15" @default.
- W4312910724 title "Sequential Decision Making with “Sequential Information” in Deep Reinforcement Learning" @default.
- W4312910724 cites W1522734439 @default.
- W4312910724 cites W2024390895 @default.
- W4312910724 cites W2086664718 @default.
- W4312910724 cites W2110485445 @default.
- W4312910724 cites W2112796928 @default.
- W4312910724 cites W2157331557 @default.
- W4312910724 cites W2168359464 @default.
- W4312910724 cites W2194775991 @default.
- W4312910724 cites W2255466643 @default.
- W4312910724 cites W2257979135 @default.
- W4312910724 cites W2424778531 @default.
- W4312910724 cites W2507009361 @default.
- W4312910724 cites W2531409750 @default.
- W4312910724 cites W2963088785 @default.
- W4312910724 cites W2963336905 @default.
- W4312910724 cites W2963524571 @default.
- W4312910724 cites W2990503944 @default.
- W4312910724 cites W3138516171 @default.
- W4312910724 cites W3175558129 @default.
- W4312910724 cites W4235411351 @default.
- W4312910724 doi "https://doi.org/10.1007/978-3-031-20868-3_13" @default.
- W4312910724 hasPublicationYear "2022" @default.
- W4312910724 type Work @default.
- W4312910724 citedByCount "0" @default.
- W4312910724 crossrefType "book-chapter" @default.
- W4312910724 hasAuthorship W4312910724A5011793745 @default.
- W4312910724 hasAuthorship W4312910724A5031014820 @default.
- W4312910724 hasAuthorship W4312910724A5051623896 @default.
- W4312910724 hasConcept C108583219 @default.
- W4312910724 hasConcept C110384440 @default.
- W4312910724 hasConcept C11413529 @default.
- W4312910724 hasConcept C115961682 @default.
- W4312910724 hasConcept C119857082 @default.
- W4312910724 hasConcept C127313418 @default.
- W4312910724 hasConcept C153180895 @default.
- W4312910724 hasConcept C154945302 @default.
- W4312910724 hasConcept C159620131 @default.
- W4312910724 hasConcept C41008148 @default.
- W4312910724 hasConcept C45374587 @default.
- W4312910724 hasConcept C62649853 @default.
- W4312910724 hasConcept C97541855 @default.
- W4312910724 hasConceptScore W4312910724C108583219 @default.
- W4312910724 hasConceptScore W4312910724C110384440 @default.
- W4312910724 hasConceptScore W4312910724C11413529 @default.
- W4312910724 hasConceptScore W4312910724C115961682 @default.
- W4312910724 hasConceptScore W4312910724C119857082 @default.
- W4312910724 hasConceptScore W4312910724C127313418 @default.
- W4312910724 hasConceptScore W4312910724C153180895 @default.
- W4312910724 hasConceptScore W4312910724C154945302 @default.
- W4312910724 hasConceptScore W4312910724C159620131 @default.
- W4312910724 hasConceptScore W4312910724C41008148 @default.
- W4312910724 hasConceptScore W4312910724C45374587 @default.
- W4312910724 hasConceptScore W4312910724C62649853 @default.
- W4312910724 hasConceptScore W4312910724C97541855 @default.
- W4312910724 hasLocation W43129107241 @default.
- W4312910724 hasOpenAccess W4312910724 @default.
- W4312910724 hasPrimaryLocation W43129107241 @default.
- W4312910724 hasRelatedWork W2773120646 @default.
- W4312910724 hasRelatedWork W3014300295 @default.
- W4312910724 hasRelatedWork W3044383684 @default.
- W4312910724 hasRelatedWork W3164822677 @default.
- W4312910724 hasRelatedWork W4223943233 @default.
- W4312910724 hasRelatedWork W4225161397 @default.
- W4312910724 hasRelatedWork W4309045103 @default.
- W4312910724 hasRelatedWork W4312200629 @default.
- W4312910724 hasRelatedWork W4360585206 @default.
- W4312910724 hasRelatedWork W4364306694 @default.
- W4312910724 isParatext "false" @default.
- W4312910724 isRetracted "false" @default.
- W4312910724 workType "book-chapter" @default.