東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Hierarchical Deep Reinforcement Lear...

University of California, Berkeley., Computer Science.

Linked to FindBook

Google Book

Amazon

博客來

Hierarchical Deep Reinforcement Learning for Robotics and Data Science.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Hierarchical Deep Reinforcement Learning for Robotics and Data Science./
Author:	Krishnan, Sanjay.
Published:	Ann Arbor : ProQuest Dissertations & Theses, : 2018,
Description:	156 p.
Notes:	Source: Dissertation Abstracts International, Volume: 80-03(E), Section: B.
Contained By:	Dissertation Abstracts International80-03B(E).
Subject:	Artificial intelligence. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10929803
ISBN:	9780438643178

Hierarchical Deep Reinforcement Learning for Robotics and Data Science.
Krishnan, Sanjay.

Hierarchical Deep Reinforcement Learning for Robotics and Data Science. - Ann Arbor : ProQuest Dissertations & Theses, 2018 - 156 p.

Source: Dissertation Abstracts International, Volume: 80-03(E), Section: B.

Thesis (Ph.D.)--University of California, Berkeley, 2018.

This dissertation explores learning important structural features of a Markov Decision Process from offline data to significantly improve the sample-efficiency, stability, and robustness of solutions even with high dimensional action spaces and long time horizons. It presents applications to surgical robot control, data cleaning, and generating efficient execution plans for relational queries. The dissertation contributes: (1) Sequential Windowed Reinforcement Learning: a framework that approximates a long-horizon MDP with a sequence of shorter term MDPs with smooth quadratic cost functions from a small number of expert demonstrations, (2) Deep Discovery of Options: an algorithm that discovers hierarchical structure in the action space from observed demonstrations, (3) AlphaClean: a system that decomposes a data cleaning task into a set of independent search problems and uses deep q-learning to share structure across the problems, and (4) Learning Query Optimizer: a system that observes executions of a dynamic program for SQL query optimization and learns a model to predict cost-to-go values to greatly speed up future search problems.

ISBN: 9780438643178Subjects--Topical Terms:

516317
Artificial intelligence.

Hierarchical Deep Reinforcement Learning for Robotics and Data Science.
LDR:02088nmm a2200289 4500 001 2201338
005 20190429062346.5
008 201008s2018 ||||||||||||||||| ||eng d
020 $a 9780438643178
035 $a (MiAaPQ)AAI10929803
035 $a (MiAaPQ)berkeley:18173
035 $a AAI10929803
040 $a MiAaPQ $c MiAaPQ
100 1 $a Krishnan, Sanjay. $3 1012585
245 1 0 $a Hierarchical Deep Reinforcement Learning for Robotics and Data Science.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2018
300 $a 156 p.
500 $a Source: Dissertation Abstracts International, Volume: 80-03(E), Section: B.
500 $a Adviser: Kenneth Goldberg.
502 $a Thesis (Ph.D.)--University of California, Berkeley, 2018.
520 $a This dissertation explores learning important structural features of a Markov Decision Process from offline data to significantly improve the sample-efficiency, stability, and robustness of solutions even with high dimensional action spaces and long time horizons. It presents applications to surgical robot control, data cleaning, and generating efficient execution plans for relational queries. The dissertation contributes: (1) Sequential Windowed Reinforcement Learning: a framework that approximates a long-horizon MDP with a sequence of shorter term MDPs with smooth quadratic cost functions from a small number of expert demonstrations, (2) Deep Discovery of Options: an algorithm that discovers hierarchical structure in the action space from observed demonstrations, (3) AlphaClean: a system that decomposes a data cleaning task into a set of independent search problems and uses deep q-learning to share structure across the problems, and (4) Learning Query Optimizer: a system that observes executions of a dynamic program for SQL query optimization and learns a model to predict cost-to-go values to greatly speed up future search problems.
590 $a School code: 0028.
650 4 $a Artificial intelligence. $3 516317
690 $a 0800
710 2 $a University of California, Berkeley. $b Computer Science. $3 1043689
773 0 $t Dissertation Abstracts International $g 80-03B(E).
790 $a 0028
791 $a Ph.D.
792 $a 2018
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10929803