Learnability of specific structural patterns of planning problems

Lukáš Chrpa, Mauro Vallati, Hugh Osborne

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

3 Citations (Scopus)

Abstract

In Automated planning, learning and exploiting additional knowledge within a domain model, in order to improve the performance of domain-independent planners, has attracted much research. Reformulation techniques such as those based on macro-operators or entanglements are very promising because they are, to some extent, domain model and planning engine independent. Despite the significant amount of work that has been done for designing techniques aimed at extracting this additional knowledge in this form, no methodological analysis has been performed for a better comprehension of their learning process. In this paper, we focus on studying learnability of entanglements in planning, in terms of how the learning process can be influenced by the quantity and the quality of the training data. So, we aim to investigate whether a small number of training planning problems is sufficient for learning a good quality set of (compatible) entanglements. Quality of the training data refers to situations where (suboptimal) plans often consist of 'flaws' (e.g. unnecessary actions). Therefore, we will investigate how the current entanglement learning approach handles such 'flaws' in training plans. Also, we will investigate whether training plans generated by different planners lead to different results of the learning process.

LanguageEnglish
Title of host publicationProceedings - 25th International Conference on Tools with Artificial Intelligence, ICTAI 2013
Pages18-23
Number of pages6
DOIs
Publication statusPublished - 2013
Event25th IEEE International Conference on Tools with Artificial Intelligence - Washington, United States
Duration: 4 Nov 20136 Nov 2013
Conference number: 25

Conference

Conference25th IEEE International Conference on Tools with Artificial Intelligence
Abbreviated titleICTAI 2013
CountryUnited States
CityWashington
Period4/11/136/11/13

Fingerprint

Planning
Defects
Macros
Engines

Cite this

Chrpa, L., Vallati, M., & Osborne, H. (2013). Learnability of specific structural patterns of planning problems. In Proceedings - 25th International Conference on Tools with Artificial Intelligence, ICTAI 2013 (pp. 18-23). [6735225] https://doi.org/10.1109/ICTAI.2013.14
Chrpa, Lukáš ; Vallati, Mauro ; Osborne, Hugh. / Learnability of specific structural patterns of planning problems. Proceedings - 25th International Conference on Tools with Artificial Intelligence, ICTAI 2013. 2013. pp. 18-23
@inproceedings{f696e48d047148778c6ad6727fe12015,
title = "Learnability of specific structural patterns of planning problems",
abstract = "In Automated planning, learning and exploiting additional knowledge within a domain model, in order to improve the performance of domain-independent planners, has attracted much research. Reformulation techniques such as those based on macro-operators or entanglements are very promising because they are, to some extent, domain model and planning engine independent. Despite the significant amount of work that has been done for designing techniques aimed at extracting this additional knowledge in this form, no methodological analysis has been performed for a better comprehension of their learning process. In this paper, we focus on studying learnability of entanglements in planning, in terms of how the learning process can be influenced by the quantity and the quality of the training data. So, we aim to investigate whether a small number of training planning problems is sufficient for learning a good quality set of (compatible) entanglements. Quality of the training data refers to situations where (suboptimal) plans often consist of 'flaws' (e.g. unnecessary actions). Therefore, we will investigate how the current entanglement learning approach handles such 'flaws' in training plans. Also, we will investigate whether training plans generated by different planners lead to different results of the learning process.",
author = "Luk{\'a}š Chrpa and Mauro Vallati and Hugh Osborne",
year = "2013",
doi = "10.1109/ICTAI.2013.14",
language = "English",
isbn = "9781479929719",
pages = "18--23",
booktitle = "Proceedings - 25th International Conference on Tools with Artificial Intelligence, ICTAI 2013",

}

Chrpa, L, Vallati, M & Osborne, H 2013, Learnability of specific structural patterns of planning problems. in Proceedings - 25th International Conference on Tools with Artificial Intelligence, ICTAI 2013., 6735225, pp. 18-23, 25th IEEE International Conference on Tools with Artificial Intelligence, Washington, United States, 4/11/13. https://doi.org/10.1109/ICTAI.2013.14

Learnability of specific structural patterns of planning problems. / Chrpa, Lukáš; Vallati, Mauro; Osborne, Hugh.

Proceedings - 25th International Conference on Tools with Artificial Intelligence, ICTAI 2013. 2013. p. 18-23 6735225.

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

TY - GEN

T1 - Learnability of specific structural patterns of planning problems

AU - Chrpa, Lukáš

AU - Vallati, Mauro

AU - Osborne, Hugh

PY - 2013

Y1 - 2013

N2 - In Automated planning, learning and exploiting additional knowledge within a domain model, in order to improve the performance of domain-independent planners, has attracted much research. Reformulation techniques such as those based on macro-operators or entanglements are very promising because they are, to some extent, domain model and planning engine independent. Despite the significant amount of work that has been done for designing techniques aimed at extracting this additional knowledge in this form, no methodological analysis has been performed for a better comprehension of their learning process. In this paper, we focus on studying learnability of entanglements in planning, in terms of how the learning process can be influenced by the quantity and the quality of the training data. So, we aim to investigate whether a small number of training planning problems is sufficient for learning a good quality set of (compatible) entanglements. Quality of the training data refers to situations where (suboptimal) plans often consist of 'flaws' (e.g. unnecessary actions). Therefore, we will investigate how the current entanglement learning approach handles such 'flaws' in training plans. Also, we will investigate whether training plans generated by different planners lead to different results of the learning process.

AB - In Automated planning, learning and exploiting additional knowledge within a domain model, in order to improve the performance of domain-independent planners, has attracted much research. Reformulation techniques such as those based on macro-operators or entanglements are very promising because they are, to some extent, domain model and planning engine independent. Despite the significant amount of work that has been done for designing techniques aimed at extracting this additional knowledge in this form, no methodological analysis has been performed for a better comprehension of their learning process. In this paper, we focus on studying learnability of entanglements in planning, in terms of how the learning process can be influenced by the quantity and the quality of the training data. So, we aim to investigate whether a small number of training planning problems is sufficient for learning a good quality set of (compatible) entanglements. Quality of the training data refers to situations where (suboptimal) plans often consist of 'flaws' (e.g. unnecessary actions). Therefore, we will investigate how the current entanglement learning approach handles such 'flaws' in training plans. Also, we will investigate whether training plans generated by different planners lead to different results of the learning process.

UR - http://www.scopus.com/inward/record.url?scp=84897713429&partnerID=8YFLogxK

U2 - 10.1109/ICTAI.2013.14

DO - 10.1109/ICTAI.2013.14

M3 - Conference contribution

SN - 9781479929719

SP - 18

EP - 23

BT - Proceedings - 25th International Conference on Tools with Artificial Intelligence, ICTAI 2013

ER -

Chrpa L, Vallati M, Osborne H. Learnability of specific structural patterns of planning problems. In Proceedings - 25th International Conference on Tools with Artificial Intelligence, ICTAI 2013. 2013. p. 18-23. 6735225 https://doi.org/10.1109/ICTAI.2013.14