Matches in SemOpenAlex for { <https://semopenalex.org/work/W2603266952> ?p ?o ?g. }
Showing items 1 to 91 of
91
with 100 items per page.
- W2603266952 abstract "We introduce the first goal-driven training for visual question answering and dialog agents. Specifically, we pose a cooperative ‘image guessing’ game between two agents – Q-BOT and A-BOT– who communicate in natural language dialog so that Q-BOT can select an unseen image from a lineup of images. We use deep reinforcement learning (RL) to learn the policies of these agents end-to-end – from pixels to multi-agent multi-round dialog to game reward.,,We demonstrate two experimental results.,,First, as a ‘sanity check’ demonstration of pure RL (from scratch), we show results on a synthetic world, where the agents communicate in ungrounded vocabularies, i.e., symbols with no pre-specified meanings (X, Y, Z). We find that two bots invent their own communication protocol and start using certain symbols to ask/answer about certain visual attributes (shape/color/style). Thus, we demonstrate the emergence of grounded language and communication among ‘visual’ dialog agents with no human supervision.,,Second, we conduct large-scale real-image experiments on the VisDial dataset [5], where we pretrain on dialog data with supervised learning (SL) and show that the RL finetuned agents significantly outperform supervised pretraining. Interestingly, the RL Q-BOT learns to ask questions that A-BOT is good at, ultimately resulting in more informative dialog and a better team." @default.
- W2603266952 created "2017-04-07" @default.
- W2603266952 creator A5014035752 @default.
- W2603266952 creator A5042265238 @default.
- W2603266952 creator A5045861415 @default.
- W2603266952 creator A5051259505 @default.
- W2603266952 creator A5070914582 @default.
- W2603266952 date "2017-10-01" @default.
- W2603266952 modified "2023-09-28" @default.
- W2603266952 title "Learning Cooperative Visual Dialog Agents with Deep Reinforcement Learning" @default.
- W2603266952 cites W1895577753 @default.
- W2603266952 cites W1895989618 @default.
- W2603266952 cites W1905882502 @default.
- W2603266952 cites W1931639407 @default.
- W2603266952 cites W1933349210 @default.
- W2603266952 cites W1947481528 @default.
- W2603266952 cites W2119717200 @default.
- W2603266952 cites W2139501017 @default.
- W2603266952 cites W2142192571 @default.
- W2603266952 cites W2251512949 @default.
- W2603266952 cites W2257979135 @default.
- W2603266952 cites W2962883855 @default.
- W2603266952 cites W2963167310 @default.
- W2603266952 cites W2963758027 @default.
- W2603266952 cites W2963890755 @default.
- W2603266952 cites W2964241990 @default.
- W2603266952 cites W2964289358 @default.
- W2603266952 cites W3089050900 @default.
- W2603266952 doi "https://doi.org/10.1109/iccv.2017.321" @default.
- W2603266952 hasPublicationYear "2017" @default.
- W2603266952 type Work @default.
- W2603266952 sameAs 2603266952 @default.
- W2603266952 citedByCount "252" @default.
- W2603266952 countsByYear W26032669522017 @default.
- W2603266952 countsByYear W26032669522018 @default.
- W2603266952 countsByYear W26032669522019 @default.
- W2603266952 countsByYear W26032669522020 @default.
- W2603266952 countsByYear W26032669522021 @default.
- W2603266952 countsByYear W26032669522022 @default.
- W2603266952 countsByYear W26032669522023 @default.
- W2603266952 crossrefType "proceedings-article" @default.
- W2603266952 hasAuthorship W2603266952A5014035752 @default.
- W2603266952 hasAuthorship W2603266952A5042265238 @default.
- W2603266952 hasAuthorship W2603266952A5045861415 @default.
- W2603266952 hasAuthorship W2603266952A5051259505 @default.
- W2603266952 hasAuthorship W2603266952A5070914582 @default.
- W2603266952 hasBestOaLocation W26032669522 @default.
- W2603266952 hasConcept C107457646 @default.
- W2603266952 hasConcept C136264566 @default.
- W2603266952 hasConcept C136764020 @default.
- W2603266952 hasConcept C154945302 @default.
- W2603266952 hasConcept C162324750 @default.
- W2603266952 hasConcept C173853756 @default.
- W2603266952 hasConcept C190954187 @default.
- W2603266952 hasConcept C195324797 @default.
- W2603266952 hasConcept C204321447 @default.
- W2603266952 hasConcept C41008148 @default.
- W2603266952 hasConcept C44291984 @default.
- W2603266952 hasConcept C90329073 @default.
- W2603266952 hasConcept C97541855 @default.
- W2603266952 hasConceptScore W2603266952C107457646 @default.
- W2603266952 hasConceptScore W2603266952C136264566 @default.
- W2603266952 hasConceptScore W2603266952C136764020 @default.
- W2603266952 hasConceptScore W2603266952C154945302 @default.
- W2603266952 hasConceptScore W2603266952C162324750 @default.
- W2603266952 hasConceptScore W2603266952C173853756 @default.
- W2603266952 hasConceptScore W2603266952C190954187 @default.
- W2603266952 hasConceptScore W2603266952C195324797 @default.
- W2603266952 hasConceptScore W2603266952C204321447 @default.
- W2603266952 hasConceptScore W2603266952C41008148 @default.
- W2603266952 hasConceptScore W2603266952C44291984 @default.
- W2603266952 hasConceptScore W2603266952C90329073 @default.
- W2603266952 hasConceptScore W2603266952C97541855 @default.
- W2603266952 hasLocation W26032669521 @default.
- W2603266952 hasLocation W26032669522 @default.
- W2603266952 hasOpenAccess W2603266952 @default.
- W2603266952 hasPrimaryLocation W26032669521 @default.
- W2603266952 hasRelatedWork W1530169779 @default.
- W2603266952 hasRelatedWork W1555369333 @default.
- W2603266952 hasRelatedWork W1607341183 @default.
- W2603266952 hasRelatedWork W1862650538 @default.
- W2603266952 hasRelatedWork W2092896632 @default.
- W2603266952 hasRelatedWork W2112259378 @default.
- W2603266952 hasRelatedWork W2174703168 @default.
- W2603266952 hasRelatedWork W2370437920 @default.
- W2603266952 hasRelatedWork W276707516 @default.
- W2603266952 hasRelatedWork W628946606 @default.
- W2603266952 isParatext "false" @default.
- W2603266952 isRetracted "false" @default.
- W2603266952 magId "2603266952" @default.
- W2603266952 workType "article" @default.