Standard
Theoretical and Experimental Results for Planning with Learned Binarized Neural Network Transition Models. / Say, Buser; Devriendt, Jo; Nordström, Jakob; Stuckey, Peter J.
Principles and Practice of Constraint Programming: 26th International Conference, CP 2020, Louvain-la-Neuve, Belgium, September 7–11, 2020, Proceedings. ed. / Helmut Simonis. Springer, 2020. p. 917-934 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12333 LNCS).
Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Harvard
Say, B, Devriendt, J
, Nordström, J & Stuckey, PJ 2020,
Theoretical and Experimental Results for Planning with Learned Binarized Neural Network Transition Models. in H Simonis (ed.),
Principles and Practice of Constraint Programming: 26th International Conference, CP 2020, Louvain-la-Neuve, Belgium, September 7–11, 2020, Proceedings. Springer, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12333 LNCS, pp. 917-934, 26th International Conference on Principles and Practice of Constraint Programming, CP 2020, Louvain-la-Neuve, Belgium,
07/09/2020.
https://doi.org/10.1007/978-3-030-58475-7_53APA
Say, B., Devriendt, J.
, Nordström, J., & Stuckey, P. J. (2020).
Theoretical and Experimental Results for Planning with Learned Binarized Neural Network Transition Models. In H. Simonis (Ed.),
Principles and Practice of Constraint Programming: 26th International Conference, CP 2020, Louvain-la-Neuve, Belgium, September 7–11, 2020, Proceedings (pp. 917-934). Springer. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) Vol. 12333 LNCS
https://doi.org/10.1007/978-3-030-58475-7_53Vancouver
Say B, Devriendt J
, Nordström J, Stuckey PJ.
Theoretical and Experimental Results for Planning with Learned Binarized Neural Network Transition Models. In Simonis H, editor, Principles and Practice of Constraint Programming: 26th International Conference, CP 2020, Louvain-la-Neuve, Belgium, September 7–11, 2020, Proceedings. Springer. 2020. p. 917-934. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12333 LNCS).
https://doi.org/10.1007/978-3-030-58475-7_53Author
Say, Buser ; Devriendt, Jo ; Nordström, Jakob ; Stuckey, Peter J. / Theoretical and Experimental Results for Planning with Learned Binarized Neural Network Transition Models. Principles and Practice of Constraint Programming: 26th International Conference, CP 2020, Louvain-la-Neuve, Belgium, September 7–11, 2020, Proceedings. editor / Helmut Simonis. Springer, 2020. pp. 917-934 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12333 LNCS).
Bibtex
@inproceedings{91b0b8ae02aa4b359efd3ecb67f88a1e,
title = "Theoretical and Experimental Results for Planning with Learned Binarized Neural Network Transition Models",
abstract = "We study planning problems where the transition function is described by a learned binarized neural network (BNN). Theoretically, we show that feasible planning with a learned BNN model is NP-complete, and present two new constraint programming models of this task as a mathematical optimization problem. Experimentally, we run solvers for constraint programming, weighted partial maximum satisfiability, 0–1 integer programming, and pseudo-Boolean optimization, and observe that the pseudo-Boolean solver outperforms previous approaches by one to two orders of magnitude. We also investigate symmetry handling for planning problems with learned BNNs over long horizons. While the results here are less clear-cut, we see that exploiting symmetries can sometimes reduce the running time of the pseudo-Boolean solver by up to three orders of magnitude.",
keywords = "Automated planning, Binarized neural networks, Cutting planes reasoning, Mathematical optimization, Pseudo-Boolean optimization, Symmetry",
author = "Buser Say and Jo Devriendt and Jakob Nordstr{\"o}m and Stuckey, {Peter J.}",
year = "2020",
doi = "10.1007/978-3-030-58475-7_53",
language = "English",
isbn = "9783030584740",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer",
pages = "917--934",
editor = "Helmut Simonis",
booktitle = "Principles and Practice of Constraint Programming",
address = "Switzerland",
note = "26th International Conference on Principles and Practice of Constraint Programming, CP 2020 ; Conference date: 07-09-2020 Through 11-09-2020",
}
RIS
TY - GEN
T1 - Theoretical and Experimental Results for Planning with Learned Binarized Neural Network Transition Models
AU - Say, Buser
AU - Devriendt, Jo
AU - Nordström, Jakob
AU - Stuckey, Peter J.
PY - 2020
Y1 - 2020
N2 - We study planning problems where the transition function is described by a learned binarized neural network (BNN). Theoretically, we show that feasible planning with a learned BNN model is NP-complete, and present two new constraint programming models of this task as a mathematical optimization problem. Experimentally, we run solvers for constraint programming, weighted partial maximum satisfiability, 0–1 integer programming, and pseudo-Boolean optimization, and observe that the pseudo-Boolean solver outperforms previous approaches by one to two orders of magnitude. We also investigate symmetry handling for planning problems with learned BNNs over long horizons. While the results here are less clear-cut, we see that exploiting symmetries can sometimes reduce the running time of the pseudo-Boolean solver by up to three orders of magnitude.
AB - We study planning problems where the transition function is described by a learned binarized neural network (BNN). Theoretically, we show that feasible planning with a learned BNN model is NP-complete, and present two new constraint programming models of this task as a mathematical optimization problem. Experimentally, we run solvers for constraint programming, weighted partial maximum satisfiability, 0–1 integer programming, and pseudo-Boolean optimization, and observe that the pseudo-Boolean solver outperforms previous approaches by one to two orders of magnitude. We also investigate symmetry handling for planning problems with learned BNNs over long horizons. While the results here are less clear-cut, we see that exploiting symmetries can sometimes reduce the running time of the pseudo-Boolean solver by up to three orders of magnitude.
KW - Automated planning
KW - Binarized neural networks
KW - Cutting planes reasoning
KW - Mathematical optimization
KW - Pseudo-Boolean optimization
KW - Symmetry
UR - http://www.scopus.com/inward/record.url?scp=85091310124&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-58475-7_53
DO - 10.1007/978-3-030-58475-7_53
M3 - Article in proceedings
AN - SCOPUS:85091310124
SN - 9783030584740
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 917
EP - 934
BT - Principles and Practice of Constraint Programming
A2 - Simonis, Helmut
PB - Springer
T2 - 26th International Conference on Principles and Practice of Constraint Programming, CP 2020
Y2 - 7 September 2020 through 11 September 2020
ER -
