Zineb Habbas | Université de Lorraine (original) (raw)

Papers by Zineb Habbas

This paper deals with solving MI-FAP problem. Because of the NP-hardness of the problem, it is di... more This paper deals with solving MI-FAP problem. Because of the NP-hardness of the problem, it is difficult to cope with real FAP instances with exact or even with heuristic methods. This paper aims at solving MI-FAP using a decomposition approach and mainly proposes a generic Top-Down approach. The key idea behind the generic aspect of our approach is to link the decomposition and the resolution steps. More precisely, two generic algorithms called Top-Down and Iterative Top-Down algorithms are proposed. To validate this approach two decomposition techniques and one efficient Adaptive Genetic Algorithm (AGA-MI-FAP) are proposed. The first results demonstrate good trade-off between the quality of solutions and the execution time.

SAT (Boolean SATisfiability Problem) is a well studied type of NP-complete problem. Most SAT solv... more SAT (Boolean SATisfiability Problem) is a well studied type of NP-complete problem. Most SAT solvers rely on software implemented tree-search based algorithms. These algorithms, basically sequential or weakly parallel, are most often ineffective when dealing with large scale instances of SAT due to the large search space to be explored in order to find one solution. Despite several improvements and heuristics being proposed for this kind of approach, the fact remains that in real problems, the computational cost continues to be prohibitive. To improve SAT solvers performance, a new trend has emerged in the late years, introducing hardware acceleration. The proposed architectures are in general hybrid, combining software and hardware parts dedicated respectively to the decisional and hard-computional parts of the algorithm. Still, most the hybrid approaches remain constrained by their limited data-access bandwith capacity. In this paper, we propose a new approach, entirely based on hardware and not depending on the SAT-instance to be solved (treated as data). It is fully configurable at synthesis in regard to the level of parallel computation and parallel buffering, as to the size of the SAT instance that can be processed (maximal number of variables and maximal clause length, the total number of clauses being unlimited). One of the main goals is to allow a more effective use of the hardware computational power by reducing the dependence to data contained in low bandwith data storage (such as RAM).

IFAC Proceedings Volumes, 2010

In this paper we propose an new heuristic criterion to determine the most critical arc of a robus... more In this paper we propose an new heuristic criterion to determine the most critical arc of a robust path in a semi dynamic graph. A robust path is defined as a collection of deviation paths covering the critical arcs. This heuristic rely on arcs fault probabilities to determine which arc of the robust path is the most critical. We experimentaly compare this heuristic with others criterions.

HAL (Le Centre pour la Communication Scientifique Directe), Apr 11, 2018

This paper presents the industrial manufacturing problem, by using the ILP (Integer Linear Progra... more This paper presents the industrial manufacturing problem, by using the ILP (Integer Linear Programming) method, which will help us to get closer to the SAT (SATisfiability) of propositional formula that is a well-known to be a NP-Complete problem. The chosen problem for this paper is the production merchandise system that has a lot of constraints, which were studied, analyzed and modeled in ILP. Our purpose for those researches is to present this industrial problem in a SAT formula, because in these last decades, a promising approach has emerged for solving efficiently large size instances by using FPGA architectures. This paper follows so this last direction and proposes a new and original way to solve them by using Integer Linear Programming, which is close to SAT.

Advances in Computational Intelligence, 2019

In this work, we investigate the use of unsupervised data mining techniques to speed up Bee Swarm... more In this work, we investigate the use of unsupervised data mining techniques to speed up Bee Swarm Optimization metaheuristic (BSO). Knowledge is extracted dynamically during the search process in order to reduce the number of candidate solutions to be evaluated. One approach uses clustering (for grouping similar solutions) and evaluates only clusters centers considered as representatives. The second uses Frequent itemset mining for guiding the search process to promising solutions. The proposed hybrid algorithms are tested on MaxSAT instances and results show that a significant reduction in time execution can be obtained for large instances while maintaining equivalent quality compared to the original BSO.

Increasing amounts of traffic, congestion, pollution and changes in the climate are reasons for a... more Increasing amounts of traffic, congestion, pollution and changes in the climate are reasons for a growing need for advanced real-time traffic and transport related information systems and services. Gathering, aggregating and ultimately transmitting this information in a practical form to the affected users remains a challenge. CARLINK (Wireless Platform for Linking Cars) has developed an intelligent wireless traffic service platform between cars supported with wireless transceivers beside the roads. WISAFECAR (Wireless Traffic Safety Platform for Linking Cars) substantively improves this platform though the addition of secured communication and data gathering and dissemination optimization. The main applications in the three participating countries (Finland, Luxembourg and South Korea) are real-time local road weather with accident warning services and dynamic urban transport service. This paper is presenting the later application as well as the overall communication and services architecture.

Advances in Computational Intelligence, 2019

Feature selection is often used before a data mining or a machine learning task in order to build... more Feature selection is often used before a data mining or a machine learning task in order to build more accurate models. It is considered as a hard optimization problem and metaheuristics give very satisfactory results for such problems. In this work, we propose a hybrid metaheuristic that integrates a reinforcement learning algorithm with Bee Swarm Optimization metaheuristic (BSO) for solving feature selection problem. QBSO-FS follows the wrapper approach. It uses a hybrid version of BSO with Q-learning for generating feature subsets and a classifier to evaluate them. The goal of using Q-learning is to benefit from the advantage of reinforcement learning to make the search process more adaptive and more efficient. The performances of QBSO-FS are evaluated on 20 well-known datasets and the results are compared with those of original BSO and other recently published methods. The results show that QBO-FS outperforms BSO-FS for large instances and gives very satisfactory results compared to recently published algorithms.

2015 27th International Conference on Microelectronics (ICM), 2015

Urban traffic congestion and bottleneck has been a global issue for many years due to rapid urban... more Urban traffic congestion and bottleneck has been a global issue for many years due to rapid urbanization. The use of intelligent Transportation System (ITS) in metropolitan cities has become a necessity. In this paper we present traffic congestion prediction models with a novel specifications based on Ant fuzzy model.

RAIRO - Operations Research, 2016

Many real-world problems can be modelled as Constraint Satisfaction Problems (CSPs). Although CSP... more Many real-world problems can be modelled as Constraint Satisfaction Problems (CSPs). Although CSP is NP-complete, it has been proved that non binary CSP instances may be efficiently solved if they are representable as Generalized Hypertree Decomposition (GHD) with small width. Algorithms for solving CSPs based on a GHD consider an extensional representation of constraints together with join and semi-join operations giving rise to new large constraint tables (or relations) needed to be temporarily saved. Extensional representation of constraints is quite natural and adapted to the specification of real problems but unfortunately limits significantly the practical performance of these algorithms. The present work tackles this problem using a compact representation of constraint tables. Consequently, to make algorithms compatible with this compact representation, new "compressed join" and "compressed semi-join" operations have to be defined. This last step constitutes our main contribution which, as far as we know, has never been presented. The correctness of this approach is proved and validated on multiple benchmarks. Experimental results show that using compressed relations and compressed operations improves significantly the practical performance of the basic algorithm proposed by Gottlob et al. for solving non binary CSPs with a Generalized Hypertree Decomposition.

Lecture Notes in Computer Science, 2016

The extraction of association rules from large transactional databases is considered in the paper... more The extraction of association rules from large transactional databases is considered in the paper using cluster architecture parallel computing. Motivated by both the successful sequential BSO-ARM algorithm, and the strong matching between this algorithm and the structure of the cluster architectures, we present in this paper a new parallel ARM algorithm that we call MW-BSO-ARM for master/worker version of BSO-ARM. The goal is to deal with large databases by minimizing the communication and synchronization costs, which represent the main challenges that faces any cluster architecture. The experimental results are very promising and show clear improvement that reaches \(300\,\%\) for large instances. For examples, in big transactional database such as WebDocs, the proposed approach generates \(10^{7}\) satisfied rules in only 22 min, while a previous GPU-based approach cannot generate more than \(10^{3}\) satisfied rules into 10 h. The results also reveal that MW-BSO-ARM outperforms the PGARM cluster-based approach in terms of computation time.

Many problems in computer science, especially in Artificial Intelligence, can be represented as c... more Many problems in computer science, especially in Artificial Intelligence, can be represented as constraint satisfaction problems (CSP). For example, scene labeling in computer vision involves testing possible interpretation of objects against relation rules. Other constraint satisfaction problems include theorem proving, scheduling, expert systems. These problems are typically NP-Complete because they require extensive searches to find a solution and the basic search algorithm is the naive Backtracking strategy. In order to improve its performances different approaches have been explored: filtering strategies, heuristics for search algorithms, decomposition methods. Although parallelization seems to be a good candidate to obtain further practical improvements the research in this direction is fewly developed. In this paper we explore the benefit of a domain decomposition strategy for parallel CSP resolution. Mainly we solve in parallel the different subproblems resulting from the de...

Association Rule Mining (ARM) is one of the most important researched techniques of data mining. ... more Association Rule Mining (ARM) is one of the most important researched techniques of data mining. The aim of ARM is to extract interesting correlations, frequent patterns, associations or causal structures among set of items in transactional databases. Many exhaustive search algorithms have been proposed for solving ARM problem. However ARM problem is algorithmically complex to deal with, especially for large data sets. For this reason, metaheuristics are increasingly considered as a more promising alternative approach for solving ARM problem. This paper follows this direction and proposes a new approach for solving ARM by using the Chemical Reaction Optimization metaheuristic(CRO). The proposed approach has been tested on two transactional datasets: the booksdataset and the food items dataset. The experimental results were compared to two state-of-the-art algorithms, namely Apriori algorithm and FP-growth algorithm. It was also compared to the binary particle swarm optimization (BPS...

Pattern Recognition and Artificial Intelligence

Association Rules Mining is an important data mining task that has many applications. Association... more Association Rules Mining is an important data mining task that has many applications. Association rules mining is considered as an optimization problem; thus several metaheuristics have been developed to solve it since they have been proven to be faster than the exact algorithms. However, most of them generates a lot of redundant rules. In this work, we proposed a modified genetic algorithm for mining interesting non-redundant association rules. Different experiments have been carried out on several well-known benchmarks. Moreover, the algorithm was compared with those of other published works and the results found proved the efficiency of our proposal.

2018 3rd International Conference on Pattern Analysis and Intelligent Systems (PAIS)

Frequent Itemsets mining is a key concept in Association Rule Mining task, it aims to discover th... more Frequent Itemsets mining is a key concept in Association Rule Mining task, it aims to discover the frequent itemsets in a transactional dataset.Nowadays large amounts of data needs to be analysed, thus the use of traditional approaches for mining frequent itemsets suffer from computational time and memory requirement, beside the difficulty for the user to provide an appropriate minimum support value. For these reasons, we propose to tackle the problem of mining frequent itemsets by mean of Chemical Reaction Optimization metaheuristic without the need to specify minsup threshold . The results show that the proposed approach gives good results and it can be used as alternative to mine frequent itemsets.

Applied Soft Computing

Abstract This paper deals with the discovery of association rules from transactional database and... more Abstract This paper deals with the discovery of association rules from transactional database and explores the combination of GPU and cluster-based parallel computing techniques. Four HPC-based bees swarm optimization approaches are proposed. The first and the second approaches called respectively BSOMW (Bees Swarm Optimization based on Master-Workers) and MWBSO (Master-Workers based on Bees Swarm Optimization) benefit from the cluster intensive computing in the generation and the evaluation steps. Given that the evaluation step is the most time consuming task, the third and fourth approaches, BSOMW-SEGPU (Bees Swarm Optimization based on Master-Workers and Single Evaluation on GPU) and MWBSO-MEGPU (Master-Workers based on Bees Swarm Optimization and Multiple Evaluation on GPU) use the GPU host in the evaluation of the generated rules. The proposed approaches are analyzed, empirically evaluated in comparison with state-of-the-art solutions. The results reveal that MWBSO-MEGPU outperforms the other proposed approaches in terms of speed up. Moreover, MWBSO-MEGPU outperforms the HPC-based ARM approaches when exploring Webdocs instance (the largest instance available on the web). The scalability of this approach is demonstrated when dealing with big transactional database (for more than 6 millions of transactions and 1 millions of items). To our knowledge, this is the first work that explores the combination of GPU and cluster-based parallel computing in association rule mining.

This paper deals with solving MI-FAP problem. Because of the NP-hardness of the problem, it is di... more This paper deals with solving MI-FAP problem. Because of the NP-hardness of the problem, it is difficult to cope with real FAP instances with exact or even with heuristic methods. This paper aims at solving MI-FAP using a decomposition approach and mainly proposes a generic Top-Down approach. The key idea behind the generic aspect of our approach is to link the decomposition and the resolution steps. More precisely, two generic algorithms called Top-Down and Iterative Top-Down algorithms are proposed. To validate this approach two decomposition techniques and one efficient Adaptive Genetic Algorithm (AGA-MI-FAP) are proposed. The first results demonstrate good trade-off between the quality of solutions and the execution time.

SAT (Boolean SATisfiability Problem) is a well studied type of NP-complete problem. Most SAT solv... more SAT (Boolean SATisfiability Problem) is a well studied type of NP-complete problem. Most SAT solvers rely on software implemented tree-search based algorithms. These algorithms, basically sequential or weakly parallel, are most often ineffective when dealing with large scale instances of SAT due to the large search space to be explored in order to find one solution. Despite several improvements and heuristics being proposed for this kind of approach, the fact remains that in real problems, the computational cost continues to be prohibitive. To improve SAT solvers performance, a new trend has emerged in the late years, introducing hardware acceleration. The proposed architectures are in general hybrid, combining software and hardware parts dedicated respectively to the decisional and hard-computional parts of the algorithm. Still, most the hybrid approaches remain constrained by their limited data-access bandwith capacity. In this paper, we propose a new approach, entirely based on hardware and not depending on the SAT-instance to be solved (treated as data). It is fully configurable at synthesis in regard to the level of parallel computation and parallel buffering, as to the size of the SAT instance that can be processed (maximal number of variables and maximal clause length, the total number of clauses being unlimited). One of the main goals is to allow a more effective use of the hardware computational power by reducing the dependence to data contained in low bandwith data storage (such as RAM).

IFAC Proceedings Volumes, 2010

In this paper we propose an new heuristic criterion to determine the most critical arc of a robus... more In this paper we propose an new heuristic criterion to determine the most critical arc of a robust path in a semi dynamic graph. A robust path is defined as a collection of deviation paths covering the critical arcs. This heuristic rely on arcs fault probabilities to determine which arc of the robust path is the most critical. We experimentaly compare this heuristic with others criterions.

HAL (Le Centre pour la Communication Scientifique Directe), Apr 11, 2018

This paper presents the industrial manufacturing problem, by using the ILP (Integer Linear Progra... more This paper presents the industrial manufacturing problem, by using the ILP (Integer Linear Programming) method, which will help us to get closer to the SAT (SATisfiability) of propositional formula that is a well-known to be a NP-Complete problem. The chosen problem for this paper is the production merchandise system that has a lot of constraints, which were studied, analyzed and modeled in ILP. Our purpose for those researches is to present this industrial problem in a SAT formula, because in these last decades, a promising approach has emerged for solving efficiently large size instances by using FPGA architectures. This paper follows so this last direction and proposes a new and original way to solve them by using Integer Linear Programming, which is close to SAT.

Advances in Computational Intelligence, 2019

In this work, we investigate the use of unsupervised data mining techniques to speed up Bee Swarm... more In this work, we investigate the use of unsupervised data mining techniques to speed up Bee Swarm Optimization metaheuristic (BSO). Knowledge is extracted dynamically during the search process in order to reduce the number of candidate solutions to be evaluated. One approach uses clustering (for grouping similar solutions) and evaluates only clusters centers considered as representatives. The second uses Frequent itemset mining for guiding the search process to promising solutions. The proposed hybrid algorithms are tested on MaxSAT instances and results show that a significant reduction in time execution can be obtained for large instances while maintaining equivalent quality compared to the original BSO.

Increasing amounts of traffic, congestion, pollution and changes in the climate are reasons for a... more Increasing amounts of traffic, congestion, pollution and changes in the climate are reasons for a growing need for advanced real-time traffic and transport related information systems and services. Gathering, aggregating and ultimately transmitting this information in a practical form to the affected users remains a challenge. CARLINK (Wireless Platform for Linking Cars) has developed an intelligent wireless traffic service platform between cars supported with wireless transceivers beside the roads. WISAFECAR (Wireless Traffic Safety Platform for Linking Cars) substantively improves this platform though the addition of secured communication and data gathering and dissemination optimization. The main applications in the three participating countries (Finland, Luxembourg and South Korea) are real-time local road weather with accident warning services and dynamic urban transport service. This paper is presenting the later application as well as the overall communication and services architecture.

Advances in Computational Intelligence, 2019

Feature selection is often used before a data mining or a machine learning task in order to build... more Feature selection is often used before a data mining or a machine learning task in order to build more accurate models. It is considered as a hard optimization problem and metaheuristics give very satisfactory results for such problems. In this work, we propose a hybrid metaheuristic that integrates a reinforcement learning algorithm with Bee Swarm Optimization metaheuristic (BSO) for solving feature selection problem. QBSO-FS follows the wrapper approach. It uses a hybrid version of BSO with Q-learning for generating feature subsets and a classifier to evaluate them. The goal of using Q-learning is to benefit from the advantage of reinforcement learning to make the search process more adaptive and more efficient. The performances of QBSO-FS are evaluated on 20 well-known datasets and the results are compared with those of original BSO and other recently published methods. The results show that QBO-FS outperforms BSO-FS for large instances and gives very satisfactory results compared to recently published algorithms.

2015 27th International Conference on Microelectronics (ICM), 2015

Urban traffic congestion and bottleneck has been a global issue for many years due to rapid urban... more Urban traffic congestion and bottleneck has been a global issue for many years due to rapid urbanization. The use of intelligent Transportation System (ITS) in metropolitan cities has become a necessity. In this paper we present traffic congestion prediction models with a novel specifications based on Ant fuzzy model.

RAIRO - Operations Research, 2016

Many real-world problems can be modelled as Constraint Satisfaction Problems (CSPs). Although CSP... more Many real-world problems can be modelled as Constraint Satisfaction Problems (CSPs). Although CSP is NP-complete, it has been proved that non binary CSP instances may be efficiently solved if they are representable as Generalized Hypertree Decomposition (GHD) with small width. Algorithms for solving CSPs based on a GHD consider an extensional representation of constraints together with join and semi-join operations giving rise to new large constraint tables (or relations) needed to be temporarily saved. Extensional representation of constraints is quite natural and adapted to the specification of real problems but unfortunately limits significantly the practical performance of these algorithms. The present work tackles this problem using a compact representation of constraint tables. Consequently, to make algorithms compatible with this compact representation, new "compressed join" and "compressed semi-join" operations have to be defined. This last step constitutes our main contribution which, as far as we know, has never been presented. The correctness of this approach is proved and validated on multiple benchmarks. Experimental results show that using compressed relations and compressed operations improves significantly the practical performance of the basic algorithm proposed by Gottlob et al. for solving non binary CSPs with a Generalized Hypertree Decomposition.

Lecture Notes in Computer Science, 2016

The extraction of association rules from large transactional databases is considered in the paper... more The extraction of association rules from large transactional databases is considered in the paper using cluster architecture parallel computing. Motivated by both the successful sequential BSO-ARM algorithm, and the strong matching between this algorithm and the structure of the cluster architectures, we present in this paper a new parallel ARM algorithm that we call MW-BSO-ARM for master/worker version of BSO-ARM. The goal is to deal with large databases by minimizing the communication and synchronization costs, which represent the main challenges that faces any cluster architecture. The experimental results are very promising and show clear improvement that reaches \(300\,\%\) for large instances. For examples, in big transactional database such as WebDocs, the proposed approach generates \(10^{7}\) satisfied rules in only 22 min, while a previous GPU-based approach cannot generate more than \(10^{3}\) satisfied rules into 10 h. The results also reveal that MW-BSO-ARM outperforms the PGARM cluster-based approach in terms of computation time.

Many problems in computer science, especially in Artificial Intelligence, can be represented as c... more Many problems in computer science, especially in Artificial Intelligence, can be represented as constraint satisfaction problems (CSP). For example, scene labeling in computer vision involves testing possible interpretation of objects against relation rules. Other constraint satisfaction problems include theorem proving, scheduling, expert systems. These problems are typically NP-Complete because they require extensive searches to find a solution and the basic search algorithm is the naive Backtracking strategy. In order to improve its performances different approaches have been explored: filtering strategies, heuristics for search algorithms, decomposition methods. Although parallelization seems to be a good candidate to obtain further practical improvements the research in this direction is fewly developed. In this paper we explore the benefit of a domain decomposition strategy for parallel CSP resolution. Mainly we solve in parallel the different subproblems resulting from the de...

Association Rule Mining (ARM) is one of the most important researched techniques of data mining. ... more Association Rule Mining (ARM) is one of the most important researched techniques of data mining. The aim of ARM is to extract interesting correlations, frequent patterns, associations or causal structures among set of items in transactional databases. Many exhaustive search algorithms have been proposed for solving ARM problem. However ARM problem is algorithmically complex to deal with, especially for large data sets. For this reason, metaheuristics are increasingly considered as a more promising alternative approach for solving ARM problem. This paper follows this direction and proposes a new approach for solving ARM by using the Chemical Reaction Optimization metaheuristic(CRO). The proposed approach has been tested on two transactional datasets: the booksdataset and the food items dataset. The experimental results were compared to two state-of-the-art algorithms, namely Apriori algorithm and FP-growth algorithm. It was also compared to the binary particle swarm optimization (BPS...

Pattern Recognition and Artificial Intelligence

Association Rules Mining is an important data mining task that has many applications. Association... more Association Rules Mining is an important data mining task that has many applications. Association rules mining is considered as an optimization problem; thus several metaheuristics have been developed to solve it since they have been proven to be faster than the exact algorithms. However, most of them generates a lot of redundant rules. In this work, we proposed a modified genetic algorithm for mining interesting non-redundant association rules. Different experiments have been carried out on several well-known benchmarks. Moreover, the algorithm was compared with those of other published works and the results found proved the efficiency of our proposal.

2018 3rd International Conference on Pattern Analysis and Intelligent Systems (PAIS)

Frequent Itemsets mining is a key concept in Association Rule Mining task, it aims to discover th... more Frequent Itemsets mining is a key concept in Association Rule Mining task, it aims to discover the frequent itemsets in a transactional dataset.Nowadays large amounts of data needs to be analysed, thus the use of traditional approaches for mining frequent itemsets suffer from computational time and memory requirement, beside the difficulty for the user to provide an appropriate minimum support value. For these reasons, we propose to tackle the problem of mining frequent itemsets by mean of Chemical Reaction Optimization metaheuristic without the need to specify minsup threshold . The results show that the proposed approach gives good results and it can be used as alternative to mine frequent itemsets.

Applied Soft Computing

Abstract This paper deals with the discovery of association rules from transactional database and... more Abstract This paper deals with the discovery of association rules from transactional database and explores the combination of GPU and cluster-based parallel computing techniques. Four HPC-based bees swarm optimization approaches are proposed. The first and the second approaches called respectively BSOMW (Bees Swarm Optimization based on Master-Workers) and MWBSO (Master-Workers based on Bees Swarm Optimization) benefit from the cluster intensive computing in the generation and the evaluation steps. Given that the evaluation step is the most time consuming task, the third and fourth approaches, BSOMW-SEGPU (Bees Swarm Optimization based on Master-Workers and Single Evaluation on GPU) and MWBSO-MEGPU (Master-Workers based on Bees Swarm Optimization and Multiple Evaluation on GPU) use the GPU host in the evaluation of the generated rules. The proposed approaches are analyzed, empirically evaluated in comparison with state-of-the-art solutions. The results reveal that MWBSO-MEGPU outperforms the other proposed approaches in terms of speed up. Moreover, MWBSO-MEGPU outperforms the HPC-based ARM approaches when exploring Webdocs instance (the largest instance available on the web). The scalability of this approach is demonstrated when dealing with big transactional database (for more than 6 millions of transactions and 1 millions of items). To our knowledge, this is the first work that explores the combination of GPU and cluster-based parallel computing in association rule mining.