The emergence of high-dimensional data requires the design of new optimization methods. Indeed, c... more The emergence of high-dimensional data requires the design of new optimization methods. Indeed, conventional optimization methods require improvements, hybridization, or parameter tuning in order to operate in spaces of high dimensions. In this paper, we present a new adaptive variant of a pattern search algorithm to solve global optimization problems exhibiting such a character. The proposed method has no parameters visible to the user and the default settings, determined by almost no a priori experimentation, are highly robust on the tested datasets. The algorithm is evaluated and compared with 11 state-of-theart methods on 20 benchmark functions of 1000 dimensions from the CEC'2010 competition. The results show that this approach obtains good performances compared to the other methods tested.
This paper proposes an enhanced Particle Swarm Optimisation (PSO) algorithm and examines its perf... more This paper proposes an enhanced Particle Swarm Optimisation (PSO) algorithm and examines its performance. In the proposed PSO approach, PSO is combined with Evolutionary Game Theory to improve convergence. One of main challenges of such stochastic optimisation algorithms is the difficulty in the theoretical analysis of the convergence and performance. Therefore, this paper analytically investigates the convergence and performance of the proposed PSO algorithm. The analysis results show that convergence speed of the proposed PSO is superior to that of the Standard PSO approach. This paper also develops another algorithm combining the proposed PSO with the Standard PSO algorithm to mitigate the potential premature convergence issue in the proposed PSO algorithm. The combined approach consists of two types of particles, one follows Standard PSO and the other follows the proposed PSO. This enables exploitation of both diversification of the particles' exploration and adaptation of the search direction.
This paper develops a novel multi-objective optimisation method based on the Evolutionary Game Th... more This paper develops a novel multi-objective optimisation method based on the Evolutionary Game Theory to solve Weapon Target Assignment problems in real-time. The main research question of this study was how to consider multi-objective functions all together and choose a best solution among many possible non-dominant optimal solutions. The key idea is the best solution can be considered as a solution which best survives in other solution spaces. Therefore, the proposed method first obtains individual solutions for each objective function. Then, Evolutionary Game Theory considers each solution as a player and evaluates them in the solution spaces of other players to check how they can survive in those spaces. The main innovation is that, unlike other multi-objective optimisation approaches, the proposed approach not only considers a set of optimal solutions regarding multiobjective functions, but also finds the best optimal solution in terms of the survivability. The stability and the real-time computation of the proposed algorithm is tested on an adapted and constrained Dynamic Weapon Target Assignment problem matching a real military requirement. The performance of the proposed approach is evaluated via numerical simulations.
Abstract: This paper suggests a simple modification to the Enhanced Unidimensional Search (EUS) m... more Abstract: This paper suggests a simple modification to the Enhanced Unidimensional Search (EUS) method where interval scanning is used to find the best value for the ratio parameter of EUS. The proposed method, called Stepped EUS (SEUS), is tested on 25 functions. The results show that in general SEUS outperforms EUS. In addition, an improved variant of SEUS, called i-restart SEUS, is also proposed and compared with 11 state-of-the-art methods. The results show that i-restart SEUS performs well, especially on multimodal ...
Engineering Applications of Artificial Intelligence, 2004
... The most widely known satellite navigation systems: are the American Global Positioning Syste... more ... The most widely known satellite navigation systems: are the American Global Positioning System (GPS), the Russian GLObal Navigation Satellite System (GLONASS) (see Fig. ... 2. Formulation a GPS surveying network as a combinatorial optimisation problem. ...
ABSTRACT Nous proposons un algorithme de sélection de caractéristiques (feature selection) à haut... more ABSTRACT Nous proposons un algorithme de sélection de caractéristiques (feature selection) à haute dimension et son application en onco-pharmacogénomique pour le cancer du sein. Dans cette application, nous devons sélectionner un ensemble de gènes dont les niveaux d'expression permettent une prédiction efficace de la réponse des patientes à un traitement de chimiothérapie préopératoire. L'algorithme de sélection de caractéristiques que nous proposons est issu d'une heuristique d'optimisation de type line search, développée pour les problèmes à grande dimensionnalité. Cet algorithme développé pour des espaces continus est ici transposé aux espaces discrets et appliqué à la sélection de sous-ensembles de sondes à ADN.
High accuracy localization is easily obtained in an exterior context with GPS, but the developmen... more High accuracy localization is easily obtained in an exterior context with GPS, but the development of indoor positioning systems remains a challenge. The GPS signals, by nature, cannot penetrate walls thus preventing this technology to provide any service indoor. This paper proposes a practical implementation of an accurate Indoor Positioning System based on Bluetooth Low Energy technology to ensure a low power, efficient and easy to setup infrastructure. It is available on any current Smartphone and requires no extra devices for the user. We use a fusion of the inertial sensors available on the device to output a precise and drift-free estimation of the user displacement, leverage this information with iBeacon radio signals used as anchors to readjust the path if needed and process all these inputs in a Particle Filter. We augmented the Fingerprinting-based likelihood calculation of the particles position’s with a unique simulation of the particle’s theoretical RSSI and also reduce...
Resources allocation and scheduling of service workflows is an important challenge in distributed... more Resources allocation and scheduling of service workflows is an important challenge in distributed computing. This is particularly true in a cloud computing environment, where many computer resources may be available at specified locations, as and when required. Quality-of-service (QoS) issues such as execution time and running costs must also be considered. Meeting this challenge requires that two classic computational problems be tackled. The first problem is allocating resources to each of the tasks in the composite web services or workflow. The second problem involves scheduling resources when each resource may be used by more than one task, and may be needed at different times. Existing approaches to scheduling workflows or composite web services in cloud computing focus only on reducing the constraint problem - such as the deadline constraint, or the cost constraint (bi-objective optimisation). This paper proposes a new genetic algorithm that solves a scheduling problem by cons...
An algorithm called Enhanced Continuous Tabu Search ECT S is proposed for the global optimization... more An algorithm called Enhanced Continuous Tabu Search ECT S is proposed for the global optimization of multiminima functions. It results from an adaptation of combinatorial Tabu Search which aims to follow, as close as possible, Glover's basic approach. In order to cover a wide domain of possi ble solutions, our algorithm first performs diversification: it locates the most promising areas, by fitting the size of the neighborhood structure to the ob jective function and its definition domain. When the most promising areas are located, the algorithm continues the search by intensification within one promis ing area of the solution space. The efficiency of ECT S is thoroughly tested by using a set of benchmark multimodal functions, of which global and local min ima are known. ECTS is compared to other published versions of continuous Tabu Search and to some alternative algorithms like Simulated Annealing.
Recent Advances on Meta-Heuristics and Their Application to Real Scenarios, 2013
Technology developments in the field of modular data links may allow the creation of a multi-link... more Technology developments in the field of modular data links may allow the creation of a multi-link communication network to be established between anti-air missiles and the launch platform. The future prospect of such ad hoc networks with many existing guidance and allocation schemes makes it possible to consider cooperative strategies for the air defence. Over the past decades, a range of missile genres have been developed mostly on the basis of one-on-one engagements which are then optimized for many-on-many scenarios. A priori allocation rules and natural missile dispersion can allow a salvo of missiles to engage a swarm of targets; however, this does not always avoid some targets leaking through the salvo, whilst other targets may experience overkill. Therefore, weapon target assignment places greater demands on the guidance chain compared with the one-on-one engagements. This study addresses a dynamic weapon target assignment (DWTA) problem in which defending weapons must protect an area against attacks from oncoming aerial threats. In order to improve the performance and efficiency of DWTA, we propose a two-step optimisation method which combines different optimisation approaches such as graph theory, evolutionary game theory (EGT), and particle swarm optimisation (PSO). The optimal DWTA problem considered in this study is to find the allocation policy enabling the area air defence against the oncoming threats in an optimal fashion. The goal is not only to destroy all the threats, but also to avoid exposing danger to own assets such as the defending weapons. Therefore, the proposed algorithm endeavours to intercept the threats as early as possible in selecting the assignment offering the more of possible date to fire. This algorithm enables to minimise the overflight inside the protected area as well.
We present two methods of feature selection in high throughput transcriptomic data, in which the ... more We present two methods of feature selection in high throughput transcriptomic data, in which the subsets of selected variables (the genes) are optima of a multi-objective function. In the clinical trials the number of embedded patient cases is never higher than in the hundreds, while the number of gene expressions measured for each patient is higher than tens of thousands. These trials aim to better understand the biology of the phenotypes at the genomic level, and to better predict the phenotypes in order to give each patient the best treatment. Our first method states that the gene subsets are the optima of a bi-objective function. This function is a tradeoff between the size of the gene subset and the discrimination of the phenotypes, expressed as the inter-class distance. Because the gene selection stage is independent of the prediction model, it is a filter method of feature selection. The second method aims to select gene subsets that will optimize the performance of a specific prediction model. It is a wrapper approach of the feature selection problem. The optimal gene subsets are computed by a line search optimization heuristic which maximizes the performances of a linear discriminant analysis. Using public datasets in oncology we compared our results to those of the main previous methods. Our optimization approach of the gene subset selection almost always returned subsets that were significantly smaller than those of the previous methods, the performance of our predictors almost always being higher, and being more robust. In the two methods we searched the space of gene subsets for optima of an explicit multi-objective function. Meta-heuristic methods are well suited to address these optimization problems, specifically in high dimensional spaces.
The emergence of high-dimensional data requires the design of new optimization methods. Indeed, c... more The emergence of high-dimensional data requires the design of new optimization methods. Indeed, conventional optimization methods require improvements, hybridization, or parameter tuning in order to operate in spaces of high dimensions. In this paper, we present a new adaptive variant of a pattern search algorithm to solve global optimization problems exhibiting such a character. The proposed method has no parameters visible to the user and the default settings, determined by almost no a priori experimentation, are highly robust on the tested datasets. The algorithm is evaluated and compared with 11 state-of-theart methods on 20 benchmark functions of 1000 dimensions from the CEC'2010 competition. The results show that this approach obtains good performances compared to the other methods tested.
This paper proposes an enhanced Particle Swarm Optimisation (PSO) algorithm and examines its perf... more This paper proposes an enhanced Particle Swarm Optimisation (PSO) algorithm and examines its performance. In the proposed PSO approach, PSO is combined with Evolutionary Game Theory to improve convergence. One of main challenges of such stochastic optimisation algorithms is the difficulty in the theoretical analysis of the convergence and performance. Therefore, this paper analytically investigates the convergence and performance of the proposed PSO algorithm. The analysis results show that convergence speed of the proposed PSO is superior to that of the Standard PSO approach. This paper also develops another algorithm combining the proposed PSO with the Standard PSO algorithm to mitigate the potential premature convergence issue in the proposed PSO algorithm. The combined approach consists of two types of particles, one follows Standard PSO and the other follows the proposed PSO. This enables exploitation of both diversification of the particles' exploration and adaptation of the search direction.
This paper develops a novel multi-objective optimisation method based on the Evolutionary Game Th... more This paper develops a novel multi-objective optimisation method based on the Evolutionary Game Theory to solve Weapon Target Assignment problems in real-time. The main research question of this study was how to consider multi-objective functions all together and choose a best solution among many possible non-dominant optimal solutions. The key idea is the best solution can be considered as a solution which best survives in other solution spaces. Therefore, the proposed method first obtains individual solutions for each objective function. Then, Evolutionary Game Theory considers each solution as a player and evaluates them in the solution spaces of other players to check how they can survive in those spaces. The main innovation is that, unlike other multi-objective optimisation approaches, the proposed approach not only considers a set of optimal solutions regarding multiobjective functions, but also finds the best optimal solution in terms of the survivability. The stability and the real-time computation of the proposed algorithm is tested on an adapted and constrained Dynamic Weapon Target Assignment problem matching a real military requirement. The performance of the proposed approach is evaluated via numerical simulations.
Abstract: This paper suggests a simple modification to the Enhanced Unidimensional Search (EUS) m... more Abstract: This paper suggests a simple modification to the Enhanced Unidimensional Search (EUS) method where interval scanning is used to find the best value for the ratio parameter of EUS. The proposed method, called Stepped EUS (SEUS), is tested on 25 functions. The results show that in general SEUS outperforms EUS. In addition, an improved variant of SEUS, called i-restart SEUS, is also proposed and compared with 11 state-of-the-art methods. The results show that i-restart SEUS performs well, especially on multimodal ...
Engineering Applications of Artificial Intelligence, 2004
... The most widely known satellite navigation systems: are the American Global Positioning Syste... more ... The most widely known satellite navigation systems: are the American Global Positioning System (GPS), the Russian GLObal Navigation Satellite System (GLONASS) (see Fig. ... 2. Formulation a GPS surveying network as a combinatorial optimisation problem. ...
ABSTRACT Nous proposons un algorithme de sélection de caractéristiques (feature selection) à haut... more ABSTRACT Nous proposons un algorithme de sélection de caractéristiques (feature selection) à haute dimension et son application en onco-pharmacogénomique pour le cancer du sein. Dans cette application, nous devons sélectionner un ensemble de gènes dont les niveaux d'expression permettent une prédiction efficace de la réponse des patientes à un traitement de chimiothérapie préopératoire. L'algorithme de sélection de caractéristiques que nous proposons est issu d'une heuristique d'optimisation de type line search, développée pour les problèmes à grande dimensionnalité. Cet algorithme développé pour des espaces continus est ici transposé aux espaces discrets et appliqué à la sélection de sous-ensembles de sondes à ADN.
High accuracy localization is easily obtained in an exterior context with GPS, but the developmen... more High accuracy localization is easily obtained in an exterior context with GPS, but the development of indoor positioning systems remains a challenge. The GPS signals, by nature, cannot penetrate walls thus preventing this technology to provide any service indoor. This paper proposes a practical implementation of an accurate Indoor Positioning System based on Bluetooth Low Energy technology to ensure a low power, efficient and easy to setup infrastructure. It is available on any current Smartphone and requires no extra devices for the user. We use a fusion of the inertial sensors available on the device to output a precise and drift-free estimation of the user displacement, leverage this information with iBeacon radio signals used as anchors to readjust the path if needed and process all these inputs in a Particle Filter. We augmented the Fingerprinting-based likelihood calculation of the particles position’s with a unique simulation of the particle’s theoretical RSSI and also reduce...
Resources allocation and scheduling of service workflows is an important challenge in distributed... more Resources allocation and scheduling of service workflows is an important challenge in distributed computing. This is particularly true in a cloud computing environment, where many computer resources may be available at specified locations, as and when required. Quality-of-service (QoS) issues such as execution time and running costs must also be considered. Meeting this challenge requires that two classic computational problems be tackled. The first problem is allocating resources to each of the tasks in the composite web services or workflow. The second problem involves scheduling resources when each resource may be used by more than one task, and may be needed at different times. Existing approaches to scheduling workflows or composite web services in cloud computing focus only on reducing the constraint problem - such as the deadline constraint, or the cost constraint (bi-objective optimisation). This paper proposes a new genetic algorithm that solves a scheduling problem by cons...
An algorithm called Enhanced Continuous Tabu Search ECT S is proposed for the global optimization... more An algorithm called Enhanced Continuous Tabu Search ECT S is proposed for the global optimization of multiminima functions. It results from an adaptation of combinatorial Tabu Search which aims to follow, as close as possible, Glover's basic approach. In order to cover a wide domain of possi ble solutions, our algorithm first performs diversification: it locates the most promising areas, by fitting the size of the neighborhood structure to the ob jective function and its definition domain. When the most promising areas are located, the algorithm continues the search by intensification within one promis ing area of the solution space. The efficiency of ECT S is thoroughly tested by using a set of benchmark multimodal functions, of which global and local min ima are known. ECTS is compared to other published versions of continuous Tabu Search and to some alternative algorithms like Simulated Annealing.
Recent Advances on Meta-Heuristics and Their Application to Real Scenarios, 2013
Technology developments in the field of modular data links may allow the creation of a multi-link... more Technology developments in the field of modular data links may allow the creation of a multi-link communication network to be established between anti-air missiles and the launch platform. The future prospect of such ad hoc networks with many existing guidance and allocation schemes makes it possible to consider cooperative strategies for the air defence. Over the past decades, a range of missile genres have been developed mostly on the basis of one-on-one engagements which are then optimized for many-on-many scenarios. A priori allocation rules and natural missile dispersion can allow a salvo of missiles to engage a swarm of targets; however, this does not always avoid some targets leaking through the salvo, whilst other targets may experience overkill. Therefore, weapon target assignment places greater demands on the guidance chain compared with the one-on-one engagements. This study addresses a dynamic weapon target assignment (DWTA) problem in which defending weapons must protect an area against attacks from oncoming aerial threats. In order to improve the performance and efficiency of DWTA, we propose a two-step optimisation method which combines different optimisation approaches such as graph theory, evolutionary game theory (EGT), and particle swarm optimisation (PSO). The optimal DWTA problem considered in this study is to find the allocation policy enabling the area air defence against the oncoming threats in an optimal fashion. The goal is not only to destroy all the threats, but also to avoid exposing danger to own assets such as the defending weapons. Therefore, the proposed algorithm endeavours to intercept the threats as early as possible in selecting the assignment offering the more of possible date to fire. This algorithm enables to minimise the overflight inside the protected area as well.
We present two methods of feature selection in high throughput transcriptomic data, in which the ... more We present two methods of feature selection in high throughput transcriptomic data, in which the subsets of selected variables (the genes) are optima of a multi-objective function. In the clinical trials the number of embedded patient cases is never higher than in the hundreds, while the number of gene expressions measured for each patient is higher than tens of thousands. These trials aim to better understand the biology of the phenotypes at the genomic level, and to better predict the phenotypes in order to give each patient the best treatment. Our first method states that the gene subsets are the optima of a bi-objective function. This function is a tradeoff between the size of the gene subset and the discrimination of the phenotypes, expressed as the inter-class distance. Because the gene selection stage is independent of the prediction model, it is a filter method of feature selection. The second method aims to select gene subsets that will optimize the performance of a specific prediction model. It is a wrapper approach of the feature selection problem. The optimal gene subsets are computed by a line search optimization heuristic which maximizes the performances of a linear discriminant analysis. Using public datasets in oncology we compared our results to those of the main previous methods. Our optimization approach of the gene subset selection almost always returned subsets that were significantly smaller than those of the previous methods, the performance of our predictors almost always being higher, and being more robust. In the two methods we searched the space of gene subsets for optima of an explicit multi-objective function. Meta-heuristic methods are well suited to address these optimization problems, specifically in high dimensional spaces.
Uploads
Papers by R. Chelouah