國立虎尾科技大學 |

Deep Reinforcement Learning in Natural Language Scenarios.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Deep Reinforcement Learning in Natural Language Scenarios./
Author:	He, Ji.
Description:	1 online resource (116 pages)
Notes:	Source: Dissertation Abstracts International, Volume: 79-04(E), Section: B.
Subject:	Artificial intelligence. -
Online resource:	click for full text (PQDT)
ISBN:	9780355355505

Deep Reinforcement Learning in Natural Language Scenarios.
He, Ji.

Deep Reinforcement Learning in Natural Language Scenarios. - 1 online resource (116 pages)

Source: Dissertation Abstracts International, Volume: 79-04(E), Section: B.

Thesis (Ph.D.)--University of Washington, 2017.

Includes bibliographical references

Reinforcement learning refers to a class of algorithms that aim at learning a good policy in a dynamic environment. Recently, by combining deep learning with reinforcement learning, researchers have made significant breakthroughs in many artificial intelligence applications. The most notable applications are Atari games and game of Go. However, natural language applications involving deep reinforcement learning are still rare.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2018

Mode of access: World Wide Web

ISBN: 9780355355505Subjects--Topical Terms:

559380
Artificial intelligence.
Index Terms--Genre/Form:

554714
Electronic books.

Deep Reinforcement Learning in Natural Language Scenarios.
LDR:03922ntm a2200361K 4500 001 912169
005 20180608102941.5
006 m o u
007 cr mn||||a|a||
008 190606s2017 xx obm 000 0 eng d
020 $a 9780355355505
035 $a (MiAaPQ)AAI10616792
035 $a (MiAaPQ)washington:17701
035 $a AAI10616792
040 $a MiAaPQ $b eng $c MiAaPQ
100 1 $a He, Ji. $3 1184405
245 1 0 $a Deep Reinforcement Learning in Natural Language Scenarios.
264 0 $c 2017
300 $a 1 online resource (116 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertation Abstracts International, Volume: 79-04(E), Section: B.
500 $a Adviser: Mari Ostendorf.
502 $a Thesis (Ph.D.)--University of Washington, 2017.
504 $a Includes bibliographical references
520 $a Reinforcement learning refers to a class of algorithms that aim at learning a good policy in a dynamic environment. Recently, by combining deep learning with reinforcement learning, researchers have made significant breakthroughs in many artificial intelligence applications. The most notable applications are Atari games and game of Go. However, natural language applications involving deep reinforcement learning are still rare.
520 $a This thesis studies deep reinforcement learning in natural language scenarios with three contributions. First we introduce a novel architecture for reinforcement learning with deep neural networks designed to handle state and action spaces characterized by natural language. The architecture represents state and action spaces with separate embedding vectors, which are combined with an interaction function to approximate the Q-function in reinforcement learning. Second, we investigate reinforcement learning with a combinatorial, natural language action space. Novel deep reinforcement learning architectures are studied for effective modeling of the value function associated with actions comprised of interdependent sub-actions, accounting for redundancy among sub-actions. In addition, a two-stage Q-learning framework is introduced as a strategy for reducing the cost to search the combinatorial action space. Third, we augment the state representation to incorporate global context using an external unstructured knowledge source with temporal information. This approach is inspired by the observation that in a real-world decision making process, it is usually beneficial to consider background knowledge and popular current events relevant to the current local context.
520 $a We experiment on two types of tasks, text-based games and predicting popular Reddit discussion threads. We show that all contributions help reinforcement learning in natural language scenarios. Specifically, experiments with paraphrased action descriptions on text games show that separate modeling of state and action spaces is extracting meaning rather than simply memorizing strings of text. For a combinatorial action space, our proposed model, which represents dependence between sub-actions through a bi-directional LSTM, gives the best performance for predicting popular Reddit threads across different domains. The two-stage Q-learning achieves significant performance gain compared to random sampling a subspace of the combinatorial action space. For tracking the most popular thread, incorporating external knowledge in the form of discussions about world news also leads to significant improvements with a 34% gain for discussions about topic (politics) for which world news is particularly relevant.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2018
538 $a Mode of access: World Wide Web
650 4 $a Artificial intelligence. $3 559380
650 4 $a Computer science. $3 573171
650 4 $a Electrical engineering. $3 596380
655 7 $a Electronic books. $2 local $3 554714
690 $a 0800
690 $a 0984
690 $a 0544
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a University of Washington. $b Electrical Engineering. $3 1180628
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10616792 $z click for full text (PQDT)