Sami Khuri

The zero/one multiple knapsack problem and genetic algorithms

A genetic algorithm, GENEsYs, is applied to an NP-complete problem, the 0/1 multiple knapsack problem. The partitioning of the search space resulting from this highly constrained problem may include substantially large infeasible regions. Our implementation allows for the breeding and participation of infeasible strings in the population. Unlike many other GA-based algorithms that are augmented with domainspecific knowledge, GENEsYs uses a simple fitness function that uses a graded penalty term to penalize infeasibly bred strings. We apply our genetic algorithm to problem instances from the literature of well known test problems and report our experimental results. These encouraging results, especially for relatively large test problems, indicate that genetic algorithms can be successfully used as heuristics for finding good solutions for highly constrained NP-complete problems.

An evolutionary approach to combinatorial optimization problems

The paper reports on the application of genetic algorithms, probabilistic search algorithms based on the model of organic evolution, to NP-complete combinatorial optimization problems. In particular, the subset sum, maximum cut, and minimum tardy task problems are considered. Except for the fitness function, no problem-specific changes of the genetic algorithm are required in order to achieve results of high quality even for the problem instances of size 100 used in the paper. For constrained problems, such as the subset sum and the minimum tardy task, the constraints are taken into account by incorporating a graded penalty term into the fitness function. Even for large instances of these highly multimodal optimization problems, an iterated application of the genetic algorithm is observed to find the global optimum within a number of runs. As the genetic algorithm samples only a tiny fraction of the search space, these results are quite encouraging.

An evolutionary heuristic for the maximum independent set problem

The results obtained from the application of a genetic algorithm, GENEsYs, to the NP-complete maximum independent set problem are reported. In contrast to many other genetic algorithm-based approaches that use domain-specific knowledge, the approach presented in this paper relies on a graded penalty term component of the fitness function to penalize infeasible solutions. The method is applied to several large problem instances of the maximum independent set problem. The results clearly indicate that genetic algorithms can be successfully used as heuristics for finding good approximative solutions for this highly constrained optimization problem.<<ETX>>

Heuristic algorithms for the terminal assignment problem

In this paper, applications of heuristic techniques for solving the terminal assignment (TA) problem are investigated. The task here is to assign terminals to concentrators in such a way that each terminal is assigned to one (and only one) concentrator and the aggregate capacity of all terminals assigned to any concentrator does not overload that concentrator, i.e., is within the concentrators capacity. Under these two hard constraints, an assignment with the lowest possible cost is sought. The proposed cost is taken to be the distance between a terminal and a concentrator. The heuristic techniques we investigate in this article include greedy-based algorithms, genetic algorithms (GA), and grouping genetic algorithms (GGA) [4]. We elaborate on the different heuristics we use, and compare the solutions yielded by them.

A Comparative Study of a Penalty Function, a Repair Heuristic and Stochastic Operators with the Set-Covering Problem

In this paper we compare the effects of using various stochastic operators with the non-unicost set-covering problem. Four different crossover operators are compared to a repair heuristic which consists in transforming infeasible strings into feasible ones. These stochastic operators are incorporated in GENEsYs, the genetic algorithm we apply to problem instances of the set-covering problem we draw from well known test problems. GENEsYs uses a simple fitness function that has a graded penalty term to penalize infeasibly bred strings. The results are compared to a non GA-based algorithm based on the greedy technique. Our computational results are then compared, shedding some light on the effects of using different operators, a penalty function, and a repair heuristic on a highly constrained combinatorial optimization problem.

An overview of visualization: its use and design: report of the working group in visualization

This paper presents an overview of visualization in Computer Science instruction. It is broken down in the following fashion. First, we present the motivation for using visualization and visual techniques in instruction. This is followed by a discussion of when the use of visualization is most appropriate. We then consider a broad spectrum of uses of visualization in Computer Science instruction. This spectrum is organized from passive to active in terms of a student’s involvement with the visualization tools. Types of visualizations are then categorized. The remainder of the paper focuses more on design issues for instructional visualization tools. These design issues are first presented from the perspective of the instructor who is constructing the visualization tool for students and then from the perspective of the programmer who is creating visualization software. We close the paper with some suggestions on organizing and maintaining a Web-based repository of visualization tools for Computer Science instruction.

Evolutionary Heuristics for the Bin Packing Problem

In this paper we investigate the use of two evolutionary based heuristic to the bin packing problem. The intractability of this problem is a motivation for the pursuit of heuristics that produce approximate solutions. Unlike other evolutionary based heuristics used with optimization problems, ours do not use domain-specific knowledge and has no specialized genetic operators. It uses a straightforward fitness function to which a graded penalty term is added to penalize infeasible strings. The encoding of the problem makes use of strings that are of integer value. Strings do not represent permutations of the objects as is the case in most approaches to this problem. We use a different representation and give justifications for our choice. Several problem instances are used with a greedy heuristic and the evolutionary based algorithms. We compare the results and conclude with some observations, and suggestions on the use of evolutionary heuristics for combinatorial optimization problems.

Interactive packages for learning image compression algorithms

This paper introduces three interactive packages for learning image compression algorithms. The first two packages, RLE and Quadtree, animate bitmap image compression algorithms, and the third package, JPEG, is a tutorial about the Joint Photographic Expert Group (JPEG) standard. The goal in designing and developing the packages was to provide instructors with tutorial and demonstration tools for teaching various interesting algorithms to students in CS1/CS2, Data Structures and Algorithms, Data Compression and Image Processing courses. The packages visualize image compression algorithms by displaying their different states of execution, using different colors to highlight the important areas, and providing textual explanations to help users understand the visualization. All three packages are interactive, platform-independent, and easy to use.

Genetic Algorithms for Solving Open Shop Scheduling Problems

In this paper we investigate the use of three evolutionary based heuristics to the open shop scheduling problem. The intractability of this problem is a motivation for the pursuit of heuristics that produce approximate solutions. This work introduces three evolutionary based heuristics, namely, a permutation genetic algorithm, a hybrid genetic algorithm and a selfish gene algorithm, and tests their applicability to the open shop scheduling problem. Several problem instances are used with our evolutionary based algorithms. We compare the results and conclude with some observations and suggestions on the use of evolutionary heuristics for scheduling problems. We also report on the success that our hybrid genetic algorithm has had on one of the large benchmark problem instances: our heuristic has produced a better solution than the current best known solution.

Animating parsing algorithms

The paper describes a package that can be used to present the parsing algorithms. The package fully animates the top-down LL(1) and bottom-up SLR(1) parsing algorithms. By full animation we mean that the input string being parsed, the corresponding actions that take place in the stack, and the building of the parse tree are all simultaneously animated on the same screen, thus enabling the user to get a full appreciation of all the intricate details that occur during parsing. The package makes use of XTANGO and can be used in the beginning of the semester as a teaching tool. Later, the students could be asked to write their own animations of the compiling process.