Abstract
To address this question, here we introduce a notion of structural similarity of plans. We conjecture that if a class of planning tasks presents structurally similar plans, then a small subset of these tasks is representative enough to learn the same knowledge (macros) as could be learnt from a larger set of tasks of the same class. We have tested our conjecture by focusing on two state-of-the-art macro generation approaches. Our large empirical analysis considering seven state-of-the-art planners, and fourteen benchmark domains from the International Planning Competition, generally confirms our conjecture which can be exploited for selecting small-yet-informative training sets of tasks.
| Original language | English |
|---|---|
| Title of host publication | 30th IEEE International Conference on Tools with Artificial Intelligence (ICTAI 2018) |
| Publisher | IEEE |
| Pages | 488-492 |
| Number of pages | 5 |
| ISBN (Electronic) | 9781538674499 |
| ISBN (Print) | 9781538674505 |
| DOIs | |
| Publication status | Published - 13 Dec 2018 |
| Event | 30th IEEE International Conference on Tools with Artificial Intelligence - Volos, Greece Duration: 5 Nov 2018 → 7 Nov 2018 Conference number: 30 http://ictai2018.org/ (Link to Conference Website) |
Conference
| Conference | 30th IEEE International Conference on Tools with Artificial Intelligence |
|---|---|
| Abbreviated title | ICTAI 2018 |
| Country | Greece |
| City | Volos |
| Period | 5/11/18 → 7/11/18 |
| Internet address |
|
Fingerprint
Cite this
}
Determining Representativeness of Training Plans : A Case of Macro-operators. / Chrpa, Lukáš; Vallati, Mauro.
30th IEEE International Conference on Tools with Artificial Intelligence (ICTAI 2018). IEEE, 2018. p. 488-492 8576079.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
TY - GEN
T1 - Determining Representativeness of Training Plans
T2 - A Case of Macro-operators
AU - Chrpa, Lukáš
AU - Vallati, Mauro
N1 - © 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
PY - 2018/12/13
Y1 - 2018/12/13
N2 - Most learning for planning approaches rely on analysis of training plans. This is especially the case for one of the best-known learning approach: the generation of macrooperators (macros). These plans, usually generated from a very limited set of training tasks, must provide a ground to extract useful knowledge that can be fruitfully exploited by planning engines. In that, training tasks have to be representative of the larger class of planning tasks on which planning engines will then be run. A pivotal question is how such a set of training tasks can be selected.To address this question, here we introduce a notion of structural similarity of plans. We conjecture that if a class of planning tasks presents structurally similar plans, then a small subset of these tasks is representative enough to learn the same knowledge (macros) as could be learnt from a larger set of tasks of the same class. We have tested our conjecture by focusing on two state-of-the-art macro generation approaches. Our large empirical analysis considering seven state-of-the-art planners, and fourteen benchmark domains from the International Planning Competition, generally confirms our conjecture which can be exploited for selecting small-yet-informative training sets of tasks.
AB - Most learning for planning approaches rely on analysis of training plans. This is especially the case for one of the best-known learning approach: the generation of macrooperators (macros). These plans, usually generated from a very limited set of training tasks, must provide a ground to extract useful knowledge that can be fruitfully exploited by planning engines. In that, training tasks have to be representative of the larger class of planning tasks on which planning engines will then be run. A pivotal question is how such a set of training tasks can be selected.To address this question, here we introduce a notion of structural similarity of plans. We conjecture that if a class of planning tasks presents structurally similar plans, then a small subset of these tasks is representative enough to learn the same knowledge (macros) as could be learnt from a larger set of tasks of the same class. We have tested our conjecture by focusing on two state-of-the-art macro generation approaches. Our large empirical analysis considering seven state-of-the-art planners, and fourteen benchmark domains from the International Planning Competition, generally confirms our conjecture which can be exploited for selecting small-yet-informative training sets of tasks.
KW - automated planning
KW - learning
KW - reformulation
KW - engines
KW - focusing
KW - benchmark testing
KW - automobiles
KW - learning (artificial intelligence)
KW - planning
KW - task analysis
KW - Training
UR - http://www.scopus.com/inward/record.url?scp=85060814036&partnerID=8YFLogxK
U2 - 10.1109/ICTAI.2018.00081
DO - 10.1109/ICTAI.2018.00081
M3 - Conference contribution
SN - 9781538674505
SP - 488
EP - 492
BT - 30th IEEE International Conference on Tools with Artificial Intelligence (ICTAI 2018)
PB - IEEE
ER -