1. A novel Hybrid Search for Minimal Perturbation Problems based
on Backjumping and Dynamic backtracking methods
EL GRAOUI EL MEHDI1
, BENELALLAM IMADE2
, BOUYAKHF EL
HOUSSINE1
1
LIMIARF, Department of Physics, Faculty of Sciences, Mohammed
V University, Rabat, Morocco
2
National Institute of Statistics and Applied Economic, Irfane
Rabat, Morocco
1www.icgst.com
http://www.icgst.com/paper.aspx?pid=P1121521384
2. Many real-life problems in Artificial Intelligence (AI) as well as in other areas can be
efficiently modeled and solved using constraint programming techniques. In many real-life
scenarios the problem is partially dynamic. For example, once a change appears in the
environment, after the original problem resolution, this change should be reflected in the
new solution. The minimal perturbation problem considers such changes, as well as the
initial solution to define a new problem whose solution should be as close as possible to the
initial solution. In this paper, we propose two new approaches: HS MPP backjumping and HS
MPP dynamic backtracking. These algorithms are based on HS MPP approach (Hybrid Search
for Minimal Perturbation Problem) [1]. They rely on the intelligent backtracking methods,
namely the backjumping and dynamic backtracking which allow reducing the number of
constraints tested and thus the computational time. The evaluation of performance is
applied for random binary problems and meeting scheduling problems, with the criteria of
computation time, number of constraints checks and number of visited nodes. Finally, the
empirical results with these search methods show the efficiency of our proposed
algorithms.
2www.icgst.com
http://www.icgst.com/paper.aspx?pid=P1121521384
A novel Hybrid Search for Minimal Perturbation Problems based on
Backjumping and Dynamic backtracking methods
Abstract
3. 3www.icgst.com
El Mehdi El Graoui received his M.Sc. in computer science and telecommunications
from Mohammed V University of Rabat, faculty of Science, Morocco in 2012. He is a
PhD student at LIMIARF Laboratory under the supervision of Mr. El Houssine
BOUYAKHF in Mohammed V University of Rabat. He has published papers in various
international conferences. His research interests include the satisfaction and
optimization of constraints problems and the artificial intelligence.
1
LIMIARF, Department of Physics, Faculty of Sciences, Mohammed V University,
Rabat, Morocco
http://www.um5a.ac.ma/index.php/en/
4. 4www.icgst.com
Imade BENELALLAM is currently an Assistant Professor teaching at the National Institute of Statistics and
applied Economic. He works also within the LIMIARF Laboratory Mohammed V University-Agdal. Imade
BENELALLAM received his Ph.D. degree in Computer Science from Mohammed V University-Agdal Morocco
in April 2010. He did his Ph.D. under the supervision of professor El Houssine Bouyakhf director of LIMIARF
Lab with the collaboration of professor Christian Bessiere director of research at CNRS, University
Montpellier 2, France. His Ph.D. research focused on Distributed Constraint Reasoning. The title of his thesis
is: “Exact approaches to DisCSPs and DCOPs problems’’. From 2005 to 2010 he was a computer science
engineer at Mohammed V University-Agdal Morocco. Now he is a member of the AI team. AI is a research
team that aims to propose and combine models and algorithms in constraints, learning and agents. He
works in different projects with industrial partner; the most significant was with Thales group. Imade
BENELALLAM is also the Chair of the GOLD Affinity Group of IEEE Morocco section. He was the IEEE GOLD
representative of Morocco section in various congresses.
National Institute of Statistics and Applied Economic, Irfane Rabat, Morocco
http://www.insea.ma/
5. 5www.icgst.com
El Houssine BOUYAKHF is full Professor at the Faculty of Sciences, Mohammed-V University, Rabat, teaching
Computer Sciences, Pattern Recognition, Image Processing and Artificial Intelligence. He is the scientific
leader of LIMIARF Lab (Laboratory of Informatics, Applied mathematics, Artificial Intelligence and Pattern
recognition). He received the Engineer degree from Sup'Aéro (ENSAE) National Higher School of Aeronautics
and Space, Toulouse, France; he received the Doctor Engineer degree in Pattern recognition and Artificial
Intelligence from University Paul Sabatier, Toulouse, France and “Doctorat d'Etat” in Robotics and Artificial
Intelligence from LAAS of CNRS and University Paul Sabatier, Toulouse, France. His main topics of interest
are: Artificial Intelligence and Constraint programming, Robotics and Vision, and Telecommunications. El
Houssine BOUYAKHF supervises several PhD theses in the research themes listed before and site leader or
key person of international projects. He has more than 100 scientific publications.
1
LIMIARF, Department of Physics, Faculty of Sciences, Mohammed V University,
Rabat, Morocco
http://www.um5a.ac.ma/index.php/en/