Search

OCR post-processing is an important step to improve OCR text accuracy. It includes two main tasks, error detection and error correction. Hill climbing algorithm is a heuristic search method used for solving optimization problems. In this paper, we present a novel OCR error correction approach using an adapted version of the Hill climbing algorithm. Correction candidates of OCR errors are explored by random character edits and evolved with the Hill climbing. The character edit patterns are obtained from the training data. The proposed model is evaluated on the benchmark dataset in the OCR post-correction competition of the International Conference on Document Analysis and Recognition 2017. It is shown that our model outperforms various baseline approaches in the competition. In addition, the randomness of the proposed algorithm is analyzed to verify its stability under parameter configurations. [ABSTRACT FROM AUTHOR]

The performed studies are aimed at the increase of the capacity of operating and prospective wheel shops by means of intensification of the use of existing equipment and creation of new specialized high-performance equipment for the repair and formation of wheelsets. To solve this problem, a technique is developed for the optimization of the parameters of equipment for wheel shops, including a mathematical model with a number of technical and technological limitations. The minimum value of the reduced costs is taken as the optimization criterion. As a result of the performed calculations, the optimal cutting conditions are obtained for the processing of carriage and locomotive axles from forgings and cross rolling products, taking into account the prospects for the development of cutting tools and the achievements of domestic and foreign machine tools. Based on the optimal ranges of changes in the rotation frequency of the principal movement and feed drive systems, and a number of other parameters of the tools, obtained as a result of calculations, the adjustments were made to the processing schemes and diagrammatic works of both existing and prospective machine tools of the enterprises wheel shops. A technical assignment is developed for the design and manufacture of a special multi-cutting semi-automatic lathe with two tool heads for axes. [ABSTRACT FROM AUTHOR]

Assembly sequence planning (ASP) plays an important role in manufacturing because it determines the investment on the assembly facilities. Recently, various research works have been done to optimize the ASP problem using different optimization algorithms. Besides the algorithm efficiency, another factor that directly contributed to the quality of the solution is how the problem is modelled. According to the existing research, the most accurate approach to represent the ASP is to model base on the part of product. However, this approach resulting in the large number of assembly sequence possibility, and finally make the optimization process become harder. This paper proposed a simpler version of part-based representation using the base part concept. In the proposed model, the base part for the product is defined prior to the establishment of the precedence constraint. This approach has contributed to a smaller search space that resulting in better chance to obtain optimum solution during the optimization. The proposed ASP modelling approach however limited the number of possible solution, since the base part was fixed prior to the optimization. [ABSTRACT FROM AUTHOR]

This paper is a continuation of the paper [8] and presents more applications of Fitzpatrick functions for solving optimization problems. The main purpose of the present work is to introduce some new properties of Fitzpatrick functions useful for solving optimization problems, using also their already presented specific properties, as the maximal monotonicity, proper, convex and lower semi-continuity. [ABSTRACT FROM AUTHOR]

This paper is a continuation of the paper [8] and presents more applications of Fitzpatrick functions for solving optimization problems. The main purpose of the present work is to introduce some new properties of Fitzpatrick functions useful for solving optimization problems, using also their already presented specific properties, as the maximal monotonicity, proper, convex and lower semi-continuity. [ABSTRACT FROM AUTHOR]

Examination timetabling is a well-studied combinatorial optimization problem involving scheduling a set of examinations into a restricted number of time-slots while satisfying a defined set of constraints. This paper presents a real-world, capacitated examination timetabling problem from Universiti Utara Malaysia (UUM), Malaysia. This dataset differs from the others reported in the literature with respect to its size, complexity and constraints. Until recently, examination timetabling in UUM is done manually with a little guidance from spreadsheet computer software for checking clashes. The propriety system is unable to do the examination timetabling automatically. Besides, the introduced datasets also consider a new constraint that has never been modeled before in timetabling literature, which is lecturer's preference. Moreover, splitting examination across several rooms and a number of hard constraints such as no mixed duration of examination within individual periods, clashes free schedule and room capacity requirement are also need to be adhered. These constraints provide an additional challenge in solving the examination timetabling problem, mainly at UUM. In this paper, graph coloring heuristics are employed to schedule examinations at each step during timetable construction. We used the concept of bin packing heuristics to assign examination to room with respect to the capacity. Since this is a new dataset and no solutions have been published in the literature yet, we only compare the results with an existing manual solution. It is found that the proposed constructive heuristic able to produce good quality solution for the tested problem. The proposed graph coloring heuristics are proved to be superior to the current method employed by the University and found to perform well in comparison. [ABSTRACT FROM AUTHOR]

Portfolio methods provide the identification of the market position of a product offered by a particular enterprise, and thus indicate the organization's long-term development prospects. In this paper the authors present the possibility of applying portfolio methods for the classification of Unmanned Aerial Vehicles (UAV) used in military engineering. The computations included 15 UAV models available on the Polish market. The analysis made it possible to classify the UAVs in question into one of the four groups, and to choose the optimal UAV model with a view to one of the criteria adopted. [ABSTRACT FROM AUTHOR]

Conjugate gradient (CG) method is an important technique in unconstrained optimization, due to its effectiveness and low memory requirements. The focus of this paper is to introduce a new CG method for solving large scale unconstrained optimization. Theoretical proofs show that the new method fulfills sufficient descent condition if strong Wolfe-Powell inexact line search is used. Besides, computational results show that our proposed method outperforms to other existing CG methods. [ABSTRACT FROM AUTHOR]

Recently, numerous studies have been concerned in conjugate gradient methods for solving large-scale unconstrained optimization method. In this paper, a three-term conjugate gradient method is proposed for unconstrained optimization which always satisfies sufficient descent direction and namely as Three-Term Rivaie-Mustafa-Ismail-Leong (TTRMIL). Under standard conditions, TTRMIL method is proved to be globally convergent under strong-Wolfe line search. Finally, numerical results are provided for the purpose of comparison. [ABSTRACT FROM AUTHOR]

The traditional approach of teaching optimization problems in calculus emphasizes more on teaching the students using analytical approach through a series of procedural steps. However, optimization normally involves problem solving in real life problems and most students fail to translate the problems into mathematic models and have difficulties to visualize the concept underlying. As an educator, it is essential to embed technology in suitable content areas to engage students in construction of meaningful learning by creating a technology-based learning environment. This paper presents the applications of technology-based learning tool in designing optimization learning activities with illustrative examples, as well as to address the challenges in the implementation of using technology in teaching and learning optimization. The suggestion activities in this paper allow flexibility for educator to modify their teaching strategy and apply technology to accommodate different level of studies for the topic of optimization. Hence, this provides great potential for a wide range of learners to enhance their understanding of the concept of optimization. [ABSTRACT FROM AUTHOR]

The field of nature inspired computing and optimization techniques have evolved to solve difficult optimization problem in diverse fields of engineering, science and technology. The firefly attraction process is mimicked in the algorithm for solving optimizing proble3ms. In Firefly Algorithm (FA) sorting of fireflies is done by using sorting algorithm. The original FA is proposed with bubble sort for ranking the fireflies. In this paper, the quick sort replaces bubble sort to decrease the time complexity of FA. The dataset used is unconstrained benchmark functions from CEC 2005 [22]. The comparison of FA using bubble sort and FA using quick sort is performed with respect to best, worst, mean, standard deviation, number of comparisons and execution time. The experimental result shows that FA using quick sort requires less number of comparisons but requires more execution time. The increased number of fireflies helps to converge into optimal solution whereas by varying dimension for algorithm performed better at a lower dimension than higher dimension. [ABSTRACT FROM AUTHOR]

The researches of periodic structures have a long history with the main contents confined in the field of forward problem. In this paper, the inverse problem is considered and an overall frame is proposed which includes two main stages, i.e., the band gap criterion and its optimization. As a preliminary investigation, the inverse problem of the flexural vibrational band gap of a periodically supported beam is analyzed. According to existing knowledge of its forward problem, the band gap criterion is given in implicit form. Then, two cases with three independent parameters, namely the double supported case and the triple one, are studied in detail and the explicit expressions of the feasible domain are constructed by numerical fitting. Finally, the parameter optimization of the double supported case with three variables is conducted using genetic algorithm aiming for the best mean attenuation within specified frequency band. [ABSTRACT FROM AUTHOR]

Conjugate gradient (CG) methods are one of the tools in optimization. Due to its low computational memory requirement, this method is used in solving several of nonlinear unconstrained optimization problems from designs, economics, physics and engineering. In this paper, a new modification of CG family coefficient (βk) is proposed and posses global convergence under exact line search direction. Numerical experimental results based on the number of iterations and central processing unit (CPU) time show that the new βk performs better than some other well known CG methods under some standard test functions. [ABSTRACT FROM AUTHOR]

The present paper describes the result of the error estimation of a numerical solution to topology optimization problems of domains in which boundary value problems of partial differential equations are defined. In the previous paper, we formulated a problem using density as a design variable, presented a regular solution by using a gradient method in a function space, and called the method the H1 gradient method. The main result in this paper shows the first order convergence in the H1 norm of the solution in the H1 gradient method with respect to the size of the finite elements if first order elements are used for the design variable and the solution to the boundary value problem. The main result is confirmed by numerical examples. [ABSTRACT FROM AUTHOR]

In the paper the problem of tool path optimization for CNC (Computer Numerical Control) cutting machines is considered. The classification of the cutting techniques is offered. We also propose a new classification of toll path problems. The tasks of cost minimization and time minimization for standard cutting technique (Continuous Cutting Problem, CCP) and for one of non-standard cutting techniques (Segment Continuous Cutting Problem, SCCP) are formalized. We show that the optimization tasks can be interpreted as discrete optimization problem (generalized travel salesman problem with additional constraints, GTSP). Formalization of some constraints for these tasks is described. For the solution GTSP we offer to use mathematical model of Prof. Chentsov based on concept of a megalopolis and dynamic programming. [ABSTRACT FROM AUTHOR]

The aim of this paper is to find an optimal individual diet that 1st. Gives all necessary ingredients with proper weighs and 2nd. Limits included unhealthy substances. Real data for some people are used and the problems are solved. [ABSTRACT FROM AUTHOR]

The search of routes with multiple Quality of Service (QoS) constraints from sources to destinations connected by networks is a Non Polynomial (NP) problem, which can be approached using Genetic Algorithms (GAs). These are heuristic algorithms that simulate the natural evolution process to solve optimization problems. In this paper we elaborate on tree-based GAs which represent the set of paths from source to destination as a tree. While existing GA applications use m-ary or binary encoding representing single paths and classes of paths in chromosomes, respectively, we propose a GA application with hybrid encoding which uses both of them. This application uses binary encoding from the first iteration until good classes of paths are found and later encodes the best paths of these classes with m-ary encoding. Its objective is taking advantage of the fast convergence due to the per class-processing of binary chromosomes and the lower computational cost of the single path processing of m-ary chromosomes. Simulation results show the performance, in terms of convergence speed, of the proposed GA application with hybrid encoding. [ABSTRACT FROM AUTHOR]

Firefly Algorithm is a recent swarm intelligence method, inspired by the social behavior of fireflies, based on their flashing and attraction characteristics [1, 2]. In this paper, we analyze the implementation of a dynamic penalty approach combined with the Firefly algorithm for solving constrained global optimization problems. In order to assess the applicability and performance of the proposed method, some benchmark problems from engineering design optimization are considered. [ABSTRACT FROM AUTHOR]

This paper proposes the impact assessment of the workers in the optimal time of operations in a Flexible Job Shop Scheduling Problem. In this work, a real enterprise was studied. The problem consists in finding the workers operations schedule, taking into account the precedence constraints. The main objective is to minimize the finish time of the last task completed in the schedule. The genetic algorithm was used to solve the optimization problem and some numerical results are presented. [ABSTRACT FROM AUTHOR]

Multi-objective optimization problems have been solved widely by determination of a Pareto optimal set. Practically, the decision-makers need to choose only one solution to implement on their system, which is a challenge for them especially when the number of solutions in the Pareto set is large. In this paper, new method has been proposed to get a good solution for multi-objective optimization problem. The method consists of two stages; the first stage used the Multi Objective Simulated Annealing algorithm to find the Pareto set that contains the non-dominated solutions, whereas the second stage used the optimal computing allocation technique to reduce the number of solutions in the Pareto set to one solution that depends on ranking the preferences of the objective functions. To validate this method, multi-objective 0\1 knapsack problem was analyzed. [ABSTRACT FROM AUTHOR]

Conjugate gradient (CG) methods represent an important computational innovation in solving large-scaled unconstrained optimization problems. There are many different versions of CG methods. Although some methods are equivalent to each other, their performances are quite different. This paper presents a new CG method based on modification of the original CG methods. The important criteria of this new CG method are its global convergence properties. Numerical result shows that this new CG method performs better than the original CG methods. [ABSTRACT FROM AUTHOR]

Conjugate gradient (CG) methods have been widely used as schemes to solve large-scale unconstrained optimization problems. Numerous studies and modifications have been done recently to improve this method. In this paper, we proposed a new type of CG coefficients (βk ) by modification of Polak and Ribiere (PR) method. This new βk is shown to possess global convergence properties by using exact line searches. Performance comparisons are made with the four most common βk proposed by the early researches. Numerical results also show that this new βk performed better. [ABSTRACT FROM AUTHOR]

Conjugate gradient methods are the most famous methods for solving nonlinear unconstrained optimization problems, especially large scale problems. That is, for its simplicity and low memory requirement. The strong Wolfe line search are usually used in practice for the analyses and implementations of conjugate gradient methods. In this paper, we present a new method of nonlinear conjugate gradient method with strong Wolfe line search for unconstrained optimization problems. Under some assumptions, the sufficient descent property and the global convergence are given. The numerical results show that our new method is efficient for some unconstrained optimization problems. [ABSTRACT FROM AUTHOR]

Conjugate gradient (CG) methods are famous for solving nonlinear unconstrained optimization problems because they only need low computer memory. In this paper, we propose a new CG coefficient (βk) which has global convergence properties using exact line search. Numerical results based on the number of iterations have shown that the new βk performs better than some other CG methods. [ABSTRACT FROM AUTHOR]

Deterministic control theory is not something new since it has been used widely in controlling the optimization of the systems. However, the assertion of stochastic framework found to be more realistic. Hence this paper discusses both deterministic and stochastic control theory and the method used respectively in their problem solving namely numerical standard methods, dynamic programming and Hamilton Jacobi-Bellman equation (HJB). Problems of funding system and its solution by using HJB equation end this study. [ABSTRACT FROM AUTHOR]

This paper presents a theory of anticipatory networks that originates from anticipatory models of consequences in multicriteria decision problems. When making a decision, the decision maker takes into account the anticipated outcomes of each future decision problem linked by the causal relations with the present one. In a network of linked decision problems, the causal relations are defined between time-ordered nodes. The scenarios of future consequences of each decision are modeled by multiple vertices starting from an appropriate node. The network is supplemented by one or more relations of anticipation, or future feedback, which describe a situation where decision makers take into account the anticipated results of some future optimization problems while making their choice. So arises a multigraph of decision problems linked causally and by one or more anticipation relation, termed here the anticipatory network. We will present the properties of anticipatory networks and propose a method of reducing, transforming and using them to solve current decision problems. Furthermore, it will be shown that most anticipatory networks can be regarded as superanticipatory systems, i.e. systems that are anticipatory in the Rosen sense and contain a future model of at least one other anticipatory system. The anticipatory networks can also be applied to filter the set of future scenarios in a foresight exercise. [ABSTRACT FROM AUTHOR]

Human-like movement is fundamental for natural human-robot interaction and collaboration. We have developed in a model for generating arm and hand movements an anthropomorphic robot. This model was inspired by the Posture-Based Motion-Planning Model of human reaching and grasping movements. In this paper we present some changes to the model we have proposed in [4] and test and compare different nonlinear constrained optimization techniques for solving the large-scale nonlinear constrained optimization problem that rises from the discretization of our time-continuous model. Furthermore, we test different time discretization steps. [ABSTRACT FROM AUTHOR]

In open vehicle routing problems, the vehicles are not required to return to the depot after completing service. In this paper, we present the first exact optimization algorithm for the open version of the well-known capacitated vehicle routing problem (CVRP). The strategy of releasing nonbasic variables from their bounds, combined with the 'active constraint' method and the notion of superbasics, has been developed for efficiently requirements; this strategy is used to force the appropriate non-integer basic variables to move to their neighborhood integer points. A study of criteria for choosing a nonbasic variable to work with in the integerizing strategy has also been made. [ABSTRACT FROM AUTHOR]

Fuzzy optimization problem has been one of the most and prominent topics inside the broad area of computational intelligent. It's especially relevant in the filed of fuzzy non-linear programming. It's application as well as practical realization can been seen in all the real world problems. In this paper a large scale non-linear fuzzy programming problem has been solved by hybrid optimization techniques of Line Search (LS), Simulated Annealing (SA) and Pattern Search (PS). As industrial production planning problem with cubic objective function, 8 decision variables and 29 constraints has been solved successfully using LS-SA-PS hybrid optimization techniques. The computational results for the objective function respect to vagueness factor and level of satisfaction has been provided in the form of 2D and 3D plots. The outcome is very promising and strongly suggests that the hybrid LS-SA-PS algorithm is very efficient and productive in solving the large scale non-linear fuzzy programming problem. [ABSTRACT FROM AUTHOR]

This paper shows a complete approach to solve a given problem, from the experimentation to the optimization of different cutting parameters. In response to an industrial problem of slotting FSX 414, a Cobalt-based refractory material, we have implemented a design of experiment to determine the most influent parameters on the tool life, the surface roughness and the cutting forces. After theses trials, an optimization approach has been implemented to find the lowest manufacturing cost while respecting the roughness constraints and cutting force limitation constraints. The optimization approach is based on the Response Surface Method (RSM) using the Sequential Quadratic programming algorithm (SQP) for a constrained problem. To avoid a local optimum and to obtain an accurate solution at low cost, an efficient strategy, which allows improving the RSM accuracy in the vicinity of the global optimum, is presented. With these models and these trials, we could apply and compare our optimization methods in order to get the lowest cost for the best quality, i.e. a satisfying surface roughness and limited cutting forces. [ABSTRACT FROM AUTHOR]

When dealing with unconstrained non-linear optimization problems using quasi-Newton methods, updating the approximation to the inverse Hessian is a computationally expensive operation and, therefore, in this paper we investigate the possibility of skipping updates on every second step. The experimental results show that the new methods (i.e. with skipping) give better performance in general than existing multi-step methods, particularly as the dimension of the test problem increases. [ABSTRACT FROM AUTHOR]