Learning probabilistic hierarchical task networks to capture user preferences

Nan Li, Subbarao Kambhampati, Sungwook Yoon

Research output: Chapter in Book/Report/Conference proceedingConference contribution

20 Citations (Scopus)

Abstract

While much work on learning in planning focused on learning domain physics (i.e., action models), and search control knowledge, little attention has been paid towards learning user preferences on desirable plans. Hierarchical task networks (HTN) are known to provide an effective way to encode user prescriptions about what constitute good plans. However, manual construction of these methods is complex and error prone. In this paper, we propose a novel approach to learning probabilistic hierarchical task networks that capture user preferences by examining user-produced plans given no prior information about the methods (in contrast, most prior work on learning within the HTN framework focused on learning "method preconditions" - i.e., domain physics - assuming that the structure of the methods is given as input). We will show that this problem has close parallels to the problem of probabilistic grammar induction, and describe how grammar inductionmethods can be adapted to learn task networks. We will empirically demonstrate the effectiveness of our approach by showing that task networks we learn are able to generate plans with a distribution close to the distribution of the userpreferred plans.

Original languageEnglish (US)
Title of host publicationIJCAI International Joint Conference on Artificial Intelligence
Pages1754-1759
Number of pages6
StatePublished - 2009
Event21st International Joint Conference on Artificial Intelligence, IJCAI-09 - Pasadena, CA, United States
Duration: Jul 11 2009Jul 17 2009

Other

Other21st International Joint Conference on Artificial Intelligence, IJCAI-09
CountryUnited States
CityPasadena, CA
Period7/11/097/17/09

Fingerprint

Physics
Planning

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Li, N., Kambhampati, S., & Yoon, S. (2009). Learning probabilistic hierarchical task networks to capture user preferences. In IJCAI International Joint Conference on Artificial Intelligence (pp. 1754-1759)

Learning probabilistic hierarchical task networks to capture user preferences. / Li, Nan; Kambhampati, Subbarao; Yoon, Sungwook.

IJCAI International Joint Conference on Artificial Intelligence. 2009. p. 1754-1759.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Li, N, Kambhampati, S & Yoon, S 2009, Learning probabilistic hierarchical task networks to capture user preferences. in IJCAI International Joint Conference on Artificial Intelligence. pp. 1754-1759, 21st International Joint Conference on Artificial Intelligence, IJCAI-09, Pasadena, CA, United States, 7/11/09.
Li N, Kambhampati S, Yoon S. Learning probabilistic hierarchical task networks to capture user preferences. In IJCAI International Joint Conference on Artificial Intelligence. 2009. p. 1754-1759
Li, Nan ; Kambhampati, Subbarao ; Yoon, Sungwook. / Learning probabilistic hierarchical task networks to capture user preferences. IJCAI International Joint Conference on Artificial Intelligence. 2009. pp. 1754-1759
@inproceedings{d41528f045044d6b92b0562d9be09f3e,
title = "Learning probabilistic hierarchical task networks to capture user preferences",
abstract = "While much work on learning in planning focused on learning domain physics (i.e., action models), and search control knowledge, little attention has been paid towards learning user preferences on desirable plans. Hierarchical task networks (HTN) are known to provide an effective way to encode user prescriptions about what constitute good plans. However, manual construction of these methods is complex and error prone. In this paper, we propose a novel approach to learning probabilistic hierarchical task networks that capture user preferences by examining user-produced plans given no prior information about the methods (in contrast, most prior work on learning within the HTN framework focused on learning {"}method preconditions{"} - i.e., domain physics - assuming that the structure of the methods is given as input). We will show that this problem has close parallels to the problem of probabilistic grammar induction, and describe how grammar inductionmethods can be adapted to learn task networks. We will empirically demonstrate the effectiveness of our approach by showing that task networks we learn are able to generate plans with a distribution close to the distribution of the userpreferred plans.",
author = "Nan Li and Subbarao Kambhampati and Sungwook Yoon",
year = "2009",
language = "English (US)",
isbn = "9781577354260",
pages = "1754--1759",
booktitle = "IJCAI International Joint Conference on Artificial Intelligence",

}

TY - GEN

T1 - Learning probabilistic hierarchical task networks to capture user preferences

AU - Li, Nan

AU - Kambhampati, Subbarao

AU - Yoon, Sungwook

PY - 2009

Y1 - 2009

N2 - While much work on learning in planning focused on learning domain physics (i.e., action models), and search control knowledge, little attention has been paid towards learning user preferences on desirable plans. Hierarchical task networks (HTN) are known to provide an effective way to encode user prescriptions about what constitute good plans. However, manual construction of these methods is complex and error prone. In this paper, we propose a novel approach to learning probabilistic hierarchical task networks that capture user preferences by examining user-produced plans given no prior information about the methods (in contrast, most prior work on learning within the HTN framework focused on learning "method preconditions" - i.e., domain physics - assuming that the structure of the methods is given as input). We will show that this problem has close parallels to the problem of probabilistic grammar induction, and describe how grammar inductionmethods can be adapted to learn task networks. We will empirically demonstrate the effectiveness of our approach by showing that task networks we learn are able to generate plans with a distribution close to the distribution of the userpreferred plans.

AB - While much work on learning in planning focused on learning domain physics (i.e., action models), and search control knowledge, little attention has been paid towards learning user preferences on desirable plans. Hierarchical task networks (HTN) are known to provide an effective way to encode user prescriptions about what constitute good plans. However, manual construction of these methods is complex and error prone. In this paper, we propose a novel approach to learning probabilistic hierarchical task networks that capture user preferences by examining user-produced plans given no prior information about the methods (in contrast, most prior work on learning within the HTN framework focused on learning "method preconditions" - i.e., domain physics - assuming that the structure of the methods is given as input). We will show that this problem has close parallels to the problem of probabilistic grammar induction, and describe how grammar inductionmethods can be adapted to learn task networks. We will empirically demonstrate the effectiveness of our approach by showing that task networks we learn are able to generate plans with a distribution close to the distribution of the userpreferred plans.

UR - http://www.scopus.com/inward/record.url?scp=78751698487&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78751698487&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:78751698487

SN - 9781577354260

SP - 1754

EP - 1759

BT - IJCAI International Joint Conference on Artificial Intelligence

ER -