Genetic programming for cross-task knowledge sharing
W. Jaskowski, K. Krawiec, and B. Wieloch. GECCO '07: Proceedings of the 9th annual conference on
Genetic and evolutionary computation, 2, page 1620--1627. London, ACM Press, (7-11 July 2007)
Abstract
We consider multi-task learning of visual concepts
within genetic programming (GP) framework. The proposed
method evolves a population of GP individuals, with
each of them composed of several GP trees that process
visual primitives derived from input images. The two
main trees are delegated to solving two different
visual tasks and are allowed to share knowledge with
each other by calling the remaining GP trees
(sub-functions) included in the same individual. The
method is applied to the visual learning task of
recognising simple shapes, using generative approach
based on visual primitives, introduced in 17. We
compare this approach to a reference method devoid of
knowledge sharing, and conclude that in the worst case
cross-task learning performs equally well, and in many
cases it leads to significant performance improvements
in one or both solved tasks.
GECCO '07: Proceedings of the 9th annual conference on
Genetic and evolutionary computation
year
2007
month
7-11 July
pages
1620--1627
publisher
ACM Press
volume
2
organisation
ACM SIGEVO (formerly ISGEC)
publisher_address
New York, NY, USA
isbn13
978-1-59593-697-4
notes
GECCO-2007 A joint meeting of the sixteenth
international conference on genetic algorithms
(ICGA-2007) and the twelfth annual genetic programming
conference (GP-2007).
ACM Order Number 910071
%0 Conference Paper
%1 1277281
%A Jaskowski, Wojciech
%A Krawiec, Krzysztof
%A Wieloch, Bartosz
%B GECCO '07: Proceedings of the 9th annual conference on
Genetic and evolutionary computation
%C London
%D 2007
%E Thierens, Dirk
%E Beyer, Hans-Georg
%E Bongard, Josh
%E Branke, Jurgen
%E Clark, John Andrew
%E Cliff, Dave
%E Congdon, Clare Bates
%E Deb, Kalyanmoy
%E Doerr, Benjamin
%E Kovacs, Tim
%E Kumar, Sanjeev
%E Miller, Julian F.
%E Moore, Jason
%E Neumann, Frank
%E Pelikan, Martin
%E Poli, Riccardo
%E Sastry, Kumara
%E Stanley, Kenneth Owen
%E Stutzle, Thomas
%E Watson, Richard A
%E Wegener, Ingo
%I ACM Press
%K algorithms, genetic knowledge learning, multitask programming, representation sharing,
%P 1620--1627
%T Genetic programming for cross-task knowledge sharing
%U http://doi.acm.org/10.1145/1276958.1277281
%V 2
%X We consider multi-task learning of visual concepts
within genetic programming (GP) framework. The proposed
method evolves a population of GP individuals, with
each of them composed of several GP trees that process
visual primitives derived from input images. The two
main trees are delegated to solving two different
visual tasks and are allowed to share knowledge with
each other by calling the remaining GP trees
(sub-functions) included in the same individual. The
method is applied to the visual learning task of
recognising simple shapes, using generative approach
based on visual primitives, introduced in 17. We
compare this approach to a reference method devoid of
knowledge sharing, and conclude that in the worst case
cross-task learning performs equally well, and in many
cases it leads to significant performance improvements
in one or both solved tasks.
@inproceedings{1277281,
abstract = {We consider multi-task learning of visual concepts
within genetic programming (GP) framework. The proposed
method evolves a population of GP individuals, with
each of them composed of several GP trees that process
visual primitives derived from input images. The two
main trees are delegated to solving two different
visual tasks and are allowed to share knowledge with
each other by calling the remaining GP trees
(sub-functions) included in the same individual. The
method is applied to the visual learning task of
recognising simple shapes, using generative approach
based on visual primitives, introduced in [17]. We
compare this approach to a reference method devoid of
knowledge sharing, and conclude that in the worst case
cross-task learning performs equally well, and in many
cases it leads to significant performance improvements
in one or both solved tasks.},
added-at = {2008-06-19T17:35:00.000+0200},
address = {London},
author = {Jaskowski, Wojciech and Krawiec, Krzysztof and Wieloch, Bartosz},
biburl = {https://www.bibsonomy.org/bibtex/288f4f2f8f78d979fa329769d835fb2c1/brazovayeye},
booktitle = {GECCO '07: Proceedings of the 9th annual conference on
Genetic and evolutionary computation},
editor = {Thierens, Dirk and Beyer, Hans-Georg and Bongard, Josh and Branke, Jurgen and Clark, John Andrew and Cliff, Dave and Congdon, Clare Bates and Deb, Kalyanmoy and Doerr, Benjamin and Kovacs, Tim and Kumar, Sanjeev and Miller, Julian F. and Moore, Jason and Neumann, Frank and Pelikan, Martin and Poli, Riccardo and Sastry, Kumara and Stanley, Kenneth Owen and Stutzle, Thomas and Watson, Richard A and Wegener, Ingo},
interhash = {fc6fd19321bbd86709934efa16dc0ff6},
intrahash = {88f4f2f8f78d979fa329769d835fb2c1},
isbn13 = {978-1-59593-697-4},
keywords = {algorithms, genetic knowledge learning, multitask programming, representation sharing,},
month = {7-11 July},
notes = {GECCO-2007 A joint meeting of the sixteenth
international conference on genetic algorithms
(ICGA-2007) and the twelfth annual genetic programming
conference (GP-2007).
ACM Order Number 910071},
organisation = {ACM SIGEVO (formerly ISGEC)},
pages = {1620--1627},
publisher = {ACM Press},
publisher_address = {New York, NY, USA},
timestamp = {2008-06-19T17:42:23.000+0200},
title = {Genetic programming for cross-task knowledge sharing},
url = {http://doi.acm.org/10.1145/1276958.1277281},
volume = 2,
year = 2007
}