Artificial intelligence enhances diagnostic accuracy and reduces subjectivity in medical imaging, especially in complex tasks like cancer detection. This study assesses several convolutional neural network architectures for classifying bladder cancer in white light cystoscopy images. Using sensitivity, specificity, and the hypervolume indicator, results show that DenseNet consistently outperforms ResNet and VGG16, offering a superior balance of diagnostic performance. These results support DenseNet’s potential for clinical use in bladder cancer diagnosis.

The Profitable Tour Problem with Simultaneous Pickup and Delivery (PTPSPD) is a practical and challenging variant of the Vehicle Routing Problem with Simultaneous Pickup and Delivery (VRPSPD), in which not all customers need to be served. This reflects real-world scenarios where fleet size is often limited and incurs significant costs. This study addresses the PTPSPD under realistic constraints and proposes an effective iterated local search algorithm that integrates complementary components to balance exploration and exploitation. A cluster-first, route-second heuristic is employed to generate high-quality initial solutions. The search process is intensified via a randomized variable neighborhood descent strategy. To enhance global exploration, a combination of random perturbation mechanisms and adaptive threshold acceptance criteria is employed. A dynamic tabu list further promotes diversification and prevents premature convergence. The proposed method is evaluated on a benchmark set of 117 PTPSPD instances involving 50 to 199 customers. Results highlight the effectiveness and robustness of the approach, establishing new best-known lower bounds for 64 instances. Additionally, the algorithm is tested on classical VRPSPD instances from the literature, further confirming its ability to consistently provide high-quality solutions within reasonable computational time.

This paper discusses a novel hybrid population-based method for scheduling multiprocessor tasks on two dedicated processors. Combining a modified grey wolf optimizer with key enhancements, it yields a robust approximate algorithm. Initialization uses a carefully selected combination of a greedy sequence and configurations, ensuring solution feasibility through a tailored sigmoid function. A straightforward local operator intensifies the search space, while a drop and rebuild operator counters premature convergence. Benchmark evaluations and comparative analysis discuss its effectiveness, highlighting the method’s ability to discover new bounds compared to recent state-of-the-art approaches.

In this paper, a population-based method, along with various local operators, is proposed to tackle the set-union knapsack problem. The devised approach integrates diverse features to form a customized cooperative method. It includes a greedy procedure for generating an initial set of solutions, an optimized greedy repair and optimization operator to handle infeasibility, a tailored local operator for exploring the search space, and the integration of a multi-neighborhood tabu search operator to enhance the global best solution. These operators are integrated into an iterative search based on the whale optimization procedure, maintaining a focus on solution quality throughout the process. Finally, an experimental study is presented to evaluate the proposed method’s performance on benchmark instances extracted from the literature. The provided results are compared with those achieved by more recent methods available in the literature, highlighting the efficiency of the method and revealing several new solutions.

This study introduces and discusses a novel hybrid discrete particle swarm optimization method tailored specifically for addressing the generalized max-mean dispersion problem. The method begins with a random initial population, which is then refined using a constructive operator. This operator iteratively selects elements based on their density, measured as the ratio of contribution to weight, thereby providing high-quality solutions. Subsequently, an adaptation of swarm optimization is employed within variable neighborhood search to perform local optimization. Within such a framework, the variable neighborhood descent method incorporates diverse neighborhood operators, including one-flip and two-flip operations, along with tabu search strategies. Moreover, to counter premature convergence, a drop and rebuild shaking strategy is integrated to explore unvisited subspaces. Empirical evidence illustrates that the designed method not only establishes new lower bounds but also matches the best-known results for other instances within reasonable time limits, highlighting its strong performance and robustness.

In this paper we propose a population-based method for tackling the k-clustering minimum bi-clique completion problem. We investigate the use of the discrete particle swarm optimisation combined with a special neighborhood decent procedure. Often configuration generated by the so-called swarm process induces infeasible solutions. In order to overcome to these situations, we introduce a tolerance strategy, where its aim is to highlight the quality of the current solution where an enlarging search space is considered. The behavior of the designed method is evaluated on some benchmark instances and its provided bounds are compared to those achieved by more recent methods available in the literature. New bounds are discovered.

In this paper, the sphere packing problem is approximately solved with a population-based method. The sphere packing problem, known as the three-dimensional knapsack, occurs in several real-world applications and because of its NP-hardness it is however computationally challenging. The designed method combines a population approach and a tolerance strategy: the population tries to maintain the diversity of a series of subsets of configurations reached throughout an iterative procedure while the tolerance strategy tries to highlight the quality of the solutions throughout the search process. The performance of the designed approach is evaluated on reference instances of the literature, where its achieved bounds are compared to those obtained by more recent algorithms.

In this article, we study the effect of the backtracking strategy when injected into solutions related to a population-based approach, especially when tackling the set-union knapsack problem. The designed method is based upon three features: (i) using a swarm optimization for generating a set of current particles, (ii) introducing an iterative search operator for providing a series of enhancing solutions linking some particles of the population and, (iii) injecting a path-relinking strategy for retrieving solutions with high quality along the paths linking two solutions from the unsearched space. The performance of the proposed method is evaluated on benchmark instances of the literature, where its achieved results are compared to those reached by the best methods available in the literature. Encouraging results have been obtained.

In this paper, we study the effect of the local branching strategy on the iterative rounding solution procedure, especially when tackling the k-clustering minimum completion problem. The designed method is based upon three features: (i) proposing an enhancing basic rounding procedure for providing a starting/current solution, (ii) dropping a part of the structure of the current solution and completing it with and extended local search combined with local branching strategy and, (iii) embedding the aforementioned phases into an iterative search. The performance of the proposed method is evaluated on benchmark instances of the literature, where its achieved results are compared to those reached by the state-of-the-art Cplex solver and the best methods available in the literature. Encouraging results have been obtained.

In this paper, we propose a hybrid particle swarm optimization-based algorithm for approximately solving the set-union knapsack problem. Knapsack problem arises in several applications such as manufacturing systems, transportation and supply chain management. This study reinforce the basic swarm optimization method with a local search procedure for the set-union knapsack problem. The proposed approach is evaluated on a benchmark instances available in the literature. The bounds achieved by the proposed approach are compared to those provided by the best methods available in the literature. Encouraging results have been provided.

This study proposes an algorithm based upon the rounding strategy for the k-clustering minimum completion problem. An instance of the problem is defined in a complete bipartite graph of S and C vertices. The goal of the problem is to decompose the initial graph into k-clusters, where each cluster is a complete bipartite subgraph. Since the problem is NP hard, any exact solver, like Cplex, is often not sufficient to achieve solutions with relatively hight quality. Thus, we propose a first alternative solution procedure for tackling large-scale instances. The designed method can be viewed as a special variant of the rounding search-based algorithm and it can be applied for solving several complex optimization problems. The proposed algorithm is evaluated on a set of benchmark instances related to the k-clustering minimum completion problem, where its achieved results are compared to the best results available in the literature.

The knapsack problem arises in real world applications, like transportation, manufacturing systems, finance, and supply chain management. In this paper, we investigate the use of a cooperative particle swarm optimization for solving the quadratic multiple knapsack problem. The standard swarm optimization is reenforced by using a local search procedure, where the swapping operator is introduced that combines items belonging to different bins (knapsacks) according to their critical items. The performance of the method is evaluated on benchmark instances of the literature, where its results are compared to the best available bounds available in the literature.

In this paper, we propose a reactive search-based algorithm for solving the problem of scheduling multiprocessor tasks on two dedicated processors.An instance of the problem is characterized by a set of tasks divided into three subsets and two processors, where some tasks can be executed either on one processor or two processors.The goal of the problem is to determine the scheduling of all tasks minimizing the execution of the last assigned task.The proposed reactive search starts with a starting greedy solution.Next, a series of local operators combined with a tabu list are introduced in order to intensify the search process.The method is also reinforced with a drop and rebuild operator that is applied for diversifying the search process.Finally, the performance of the proposed method is evaluated on a set of benchmark instances, where its provided results are compared to those achieved by a recent method available in the literature.Encouraging results have been reached.

9th International Conference on Operations Research and Enterprise Systems (ICORES), Valletta, MALTA, FEB 22-24, 2020

In this paper, we propose an reactive method to solve the Quadratic Knapsack Problem (noted QKP). The quadratic knapsack problem is a well-studied combinatorial optimisation problem. In all variants of the quadratic knapsack problems, for a set of given items, profits are not only assigned to individual items but also to pairs of them. The pairwise profit is added to the quadratic objective value only when the two corresponding items are both included in the same knapsack. We let a knapsack with a capacity C and a set of candidate objects (or items), each item has a positive weight wi, and profit Pi, if selected, generates an object profit piand a pairwise profit Pijwith any other selected object j. The objective of the QKP is to select a subset of objects to fill the knapsack so as to maximize the overall profit P, while the overall weight does not exceed the knapsack capacity C. This problem is known as quadratic knapsack problem and has been shown strongly NP-hard and it has been applied with a variety of important applications, such as, in the location of satellites, airports, railway stations or freight terminals. The proposed method is mainly based on two complementary phases. In the first phase we use a greedy algorithm for provide the starting solution. In the second phase we improved the quality of the starting solutions based on a reactive search with applying a destroy and repair process, the reactive phase is based both diversification and intensification strategy. The obtained results are compared to those reached by the Cplex solverl and literature. Consequently, the experimental results show the effectiveness of the proposed approach, and the computational results show that the algorithm is capable of solving instances of the QKP that cannot be solved by other methods.

The Team Orienteering Problem (namely TOP) consists in finding the routings for a set of vehicles that maximize the total profit reached by visiting a series of customers. In this paper, a hybrid multi-objective evolutionary algorithm based on a special genetic algorithm and local search operators is proposed for approximately solving the TOP. Two conflicting objectives are considered: to minimize the total travel cost and to maximize the profit linked to the visited customers. The performance of the proposed method is evaluated on a set of benchmark instances extracted from Chao et al. [2] and its provided results are compared to those reached by the best methods available in the literature. Encouraging results have been obtained.

In this paper, the identical sphere packing is tackled by applying a particle swarm optimization-based method. An instance of the problem is characterized by a set of equal spheres and a large sphere with unlimited radius. The aim of the problem is to determine a minimum radius of the spherical container that contains all spheres without overlapping. The particle swarm optimization cooperates with an efficient continuous local optimization that serves either to repair the non-feasibility of solutions or improve their quality. The behavior of the proposed method is evaluated on a set of standard benchmark instances taken from the literature and its achieved results are compared to those obtained by the best methods available in the literature. As shown in the experimental part, the proposed approach is very competitive.

In this paper, we propose the neighborhood search-based heuristic to solve the k-clustering minimum biclique completion problem. The KCmBCP consists in partitioning a bipartite undirected graph into k clusters such that the total number of edges that added for complete each cluster into a biclique is minimum. Given a bipartite graph G = (S, T, E), we consider the problem of finding k bipartite sub-graphs, called clusters, such that each vertex i of S appears in exactly one of them, every vertex j of T appears in each cluster in which at least one of its neighbors appears, and the total number of edges needed to make each cluster complete is minimized. This problem is known as k-clustering minimum biclique completion problem (noted KCmBCP) and has been shown strongly NP-hard. The proposed approach begins by construction of an initial solution. After, the proposed approach enters on iterative phase, which destroys part of the initial solution and build of a feasible solution by exploring the neighborhood of the current solution. The experimental results show the effectiveness of the proposed approach.

In this paper, we propose a diversified method for tackling the multi-scenarios max-min knapsack problem. The proposed method is based on three phases: (i) the building phase, (ii) the combination phase and (iii) the exploring phase. The first phase yields a feasible solution by using a greedy procedure. The second phase tries to provide a new solution by combining subsets of starting solutions. The third phase tries to make an intensification in order to improve the solutions at hand. The proposed method is evaluated on a set of benchmark instances taken from the literature. The obtained results are compared to those reached by the best algorithms available in the literature. The results show that the proposed method provides better solutions than those already published.

In this paper we investigate the use of the large neighborhood search for solving the vehicle routing problem with two-dimensional loading constraints, an NP-hard combinatorial optimization problem. Such a problem may be viewed as the combination of two complementary well-known problems: two-dimensional bin-packing and capacitated vehicle routing. The proposed method considers a two-phase solution procedure: a first phase generates a placement order and applies a bottom-left strategy to this order for packing the items in the vehicle and, a second phase is applied for reaching a feasible routing. Both phases are combined by using a diversification strategy. The proposed method has been evaluated on Gendreau's benchmark instances. The obtained results are compared to those reached by the best methods taken from the literature. Encouraging results have been obtained.

In this paper, we propose to solve the max-min multiple knapsack problem by using an exact solution search. An instance of the problem is defined by a set of n items to be packed into m knapsacks as to maximize the minimum of the knapsacks' profits. The proposed method uses a series of interval searches, where each interval is bounded with a target value (considered as a lower bound) and an estimated upper bound. The target lower bound is computed by applying some aggressive fixation of some items to optimality whereas the upper bound is computed by using a surrogate relaxation. The performance of the proposed method is evaluated on a set of instances containing a variety of sizes. Computational results showed the superiority of the proposed method when comparing its provided results to those obtained by the Cplex solver and one of the best exact method available in the literature.

In this paper, we investigate the use of the large neighborhood search for solving the pickup and delivery problem with time windows. Such a problem may be viewed as a variant of the capacitated vehicle routing problem with time windows, where both precedence and coupling constraints are considered. The proposed method is based on the framework of the large neighborhood search combined with local search procedures. In order to evaluate the performance of the proposed method, it has been tested on Li et al. 's benchmark instances. The obtained results are compared to those reached by the best method available in the literature. Encouraging results have been obtained.

In this paper we propose to solve the Vehicle Routing Problem with Time Windows (VRPTW) using a hybrid metaheuristic. The VRPTW is a bi-objective optimization problem where both the number of vehicles and the distance of the travel to use should be minimized. Because it is often difficult to optimize both objectives, we propose an approach that optimizes the distance traveled by a fleet of vehicles. Such a strategy has been already used by several authors in the domain. Herein, an instance of VRPTW is considered as the composition of the Assignment Problem and a series of Traveling Salesman Problems with Time Windows (TSPTW). Both AP and TSPTW are solved by using an ant colony optimization system. Furthermore, in order to enhance the quality of the current solution, a large neighborhood search is introduced. Finally, a preliminary experimental part is presented where the proposed method is evaluated on a set of benchmark instances and its results are compared to the best results obtained by the methods available in the literature. Our preliminary results show that the proposed hybrid method remains competitive and it is able to reach new minimum distances for some tested instances.

In this paper we propose to enhance a width-beam search in order to solve the three-dimensional sphere packing problem. The goal of the problem is to determine the minimum length of the container having fixed width and height, that packs n predefined unequal spheres. The width-beam search uses a greedy selection phase which determines a subset of eligible positions for packing the predefined items in the target object and selects a subset of nodes for exploring some promising paths. We propose to handle lower bounds in the tree and apply a hill-climbing strategy in order to diversify the search process. The performance of the proposed method is evaluated on benchmark instances taken from the literature. The obtained results are compared to those reached by some recent methods available in the literature. Encouraging results have been obtained.

This paper proposes a parallel peer-to-peer cooperative algorithm for approximately solving the constrained fixed-orientation two-staged two-dimensional cutting problem. The resolution process is based in three mechanisms: a beam-search search strategy, a strip generation filling procedure, and a upper bound applied for refining the selected paths. The algorithm explores, in parallel, a subset of elite nodes where each processor develops its own path according to its internal lists. the algorithms adapts to the number of processors available on the peer-to-peer platform by backuping the plateform the partial solutions. The computational investigation on the P2Pdc environment shows the good efficiency of the proposed algorithm.

This paper solves FC_2TDC, the two-staged fixed-orientation two-dimensional cutting stock problem. FC_2TDC is a variant of the constrained two-dimensional cutting problem where each piece is produced, in the final cutting pattern, by at most two cuts, and each piece has a fixed orientation. It is solved by an algorithm that combines a strip-generation procedure with beam search (BS). BS, which is a truncated branch-and-bound, investigates a subset of elite nodes and permanently fathoms the others. It chooses the elite nodes based on an evaluation operator. The importance of the evaluation operator on the performance of BS is highlighted by comparing a local beam search (LBS) that uses a priority cost-evaluation operator to a global beam search (GBS) that adopts a global cost-evaluation operator. Computational results, based on several medium and large instances, further show that LBS provides good solutions, whereas GBS gives (near) optimal solutions within reasonable computational time.